detecting_loanwords/loanwords_problems_solutions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The Problem of Loanwords: Detection and Remedies"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What is a Loanword?\n",
    "They are real: https://en.wikipedia.org/wiki/Loanword but they often overlooked in everyday use. Take this example conversation:\n",
    "* \"It was nice seeing you again. Adios amigo.\"\n",
    "* Until we meet again, \"Vaya con Dios!\"\n",
    "\n",
    "#### The last two sentences are loanwords, borrowings from Spanish. Most participants can navigate and interpret conversations even if they don't know the exact translations of the loanwords.\n",
    "\n",
    "However, when doing Natural Language Processing loanwords may become a pesky problem, for example:\n",
    "* If you are trying to generate a high quality embedding\n",
    "    * you many need to detect and drop out the loanwords since they are out of scope for typical monolingual embeddings.\n",
    "* If your application is translating a sequence of sentences\n",
    "    * you may need to switch models to create a better translation of a loanword sequence.\n",
    "* Some loanwords may occur in both languages \n",
    "    * you will need to use other heuristic measures to increase your precision and recall in detection.\n",
    "#### etc.\n",
    "### So let's look at how we can detect and work around these problem areas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### In this notebook you will:\n",
    "1. Perform ETL (Extract-Transform-Load)\n",
    "    1. Extract data using a Corpus Reader object\n",
    "    1. Transform the data using a reusable, composable Scikit-Learn Pipeline object\n",
    "    1. Load the data into a matrix and train a classifier\n",
    "1. Train several classifiers and select the best algorithm for the data\n",
    "1. Use GridSearch to tune hyperparameters to achieve the best algorithm performance\n",
    "1. Use the classifier to predict the classification of each word\n",
    "1. Extract whole phrases by clustering occurences of predicted words \n",
    "1. Save the pipelines and classifier for reuse at runtime\n",
    "1. Record the trained classifier model's provenance\n",
    "1. Examine some unseen, untrained data to discover how well the classifier generalizes to unseen data\n",
    "\n",
    "## Our problem: Detect Transliterated Greek in Classical Latin Authors\n",
    "### Why? \n",
    "### To make a high quality word embedding, using a relatively limited corpus, it's important to filter out foreign words. Often Latin authors will quote Greek authors and transliterate the Greek into Latin equivalents; however the resulting words often haven't been generally adopted in the language. True, some transliterated Greek words are valid Latin words, but we aren't concerned about a word here or there, but rather clusters of foreign words in the source language. \n",
    "### We will generate our data sets using CLTK (Classical Language Toolkit http://cltk.org/ ):\n",
    "* The Latin Library\n",
    "    * Julius Caesar\n",
    "    * Prudentius\n",
    "    * Eutropius\n",
    "* The Perseus Library Greek Texts\n",
    "    * The Works of Plato\n",
    "    * Homer's Odyssey\n",
    "* We will augment our Latin data set using probability distributions in tandem with CLTK's lemmatization dictionary.\n",
    "\n",
    "#### The probability distributions come from our other notebooks:\n",
    "* `probablistic_language_modeling/make_frequency_distribution.ipynb` \n",
    "* `detecting_loanwords/make_frequency_distribution_greek_transliterated.ipynb`\n",
    "\n",
    "#### The Greek will be transliterated into Latin. \n",
    "#### We will use our classifier to examine the corpus of Pliny the Younger to detect the use of transliterated Greek words, and we'll assess the classifier's effectiveness on the entire Latin Library corpus."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime\n",
    "import glob\n",
    "import json\n",
    "import logging\n",
    "import multiprocessing\n",
    "import os\n",
    "import site\n",
    "from copy import deepcopy\n",
    "import random\n",
    "from collections import defaultdict\n",
    "import pickle\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import learning_curve, ShuffleSplit, train_test_split, GridSearchCV, cross_val_score\n",
    "from scipy import sparse\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import BaggingClassifier, ExtraTreesClassifier, RandomForestClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.utils.extmath import density\n",
    "from sklearn.dummy import DummyClassifier\n",
    "from joblib import dump, load\n",
    "import sklearn\n",
    "from sklearn.preprocessing import FunctionTransformer, LabelEncoder, StandardScaler\n",
    "from cltk.stem.latin.j_v import JVReplacer\n",
    "from cltk.prosody.latin.scansion_constants import ScansionConstants\n",
    "from cltk.prosody.latin.string_utils import remove_punctuation_dict\n",
    "from cltk.tokenize.word import WordTokenizer\n",
    "from cltk.corpus.readers import get_corpus_reader\n",
    "from cltk.utils.featurization import word_to_features\n",
    "from cltk.utils.file_operations import md5\n",
    "from cltk.utils.matrix_corpus_fun import (\n",
    "    distinct_words,\n",
    "    separate_camel_cases,\n",
    "    drop_empty_lists,\n",
    "    drop_non_lower,\n",
    "    drop_arabic_numeric,\n",
    "    drop_all_caps,\n",
    "    drop_empty_strings,\n",
    "    jv_transform,\n",
    "    splice_hyphens,\n",
    "    accept_editorial,\n",
    "    profile_chars,\n",
    "    demacronize,\n",
    "    drop_enclitics,\n",
    "    drop_fringe_punctuation,\n",
    "    divide_separate_words,\n",
    "    drop_all_punctuation)\n",
    "plt.style.use('fivethirtyeight')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add parent directory to path so we can access our common code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "import inspect\n",
    "currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))\n",
    "parentdir = os.path.dirname(currentdir)\n",
    "sys.path.insert(0, parentdir) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mlyoucanuse.romanizer import Romanizer, romanizer_transform  \n",
    "from mlyoucanuse.aeoe_replacer import aeoe_transform\n",
    "from mlyoucanuse.matrix_fun import (run_length_encoding,\n",
    "                                    extract_words,                                     \n",
    "                                    patch_cluster_holes)\n",
    "from mlyoucanuse.featurize_text_fun import word_to_features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Turn on logging, primarily so that library methods may report warnings, if any"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "LOG = logging.getLogger('make_model')\n",
    "LOG.addHandler(logging.NullHandler())\n",
    "logging.basicConfig(level=logging.INFO)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define a CorpusReader and select only the text files of Prudentius, Caesar and Eutropius. \n",
    "#### As shown in the appendix of this notebook, the authors of this seed set have a low incidence of using transliterated Greek words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:make_model:available good files 41\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['caesar/alex.txt',\n",
       " 'caesar/bc1.txt',\n",
       " 'caesar/bc2.txt',\n",
       " 'caesar/bc3.txt',\n",
       " 'caesar/bellafr.txt',\n",
       " 'caesar/gall1.txt',\n",
       " 'caesar/gall2.txt',\n",
       " 'caesar/gall3.txt',\n",
       " 'caesar/gall4.txt',\n",
       " 'caesar/gall5.txt',\n",
       " 'caesar/gall6.txt',\n",
       " 'caesar/gall7.txt',\n",
       " 'caesar/gall8.txt',\n",
       " 'caesar/hisp.txt',\n",
       " 'eutropius/eutropius1.txt',\n",
       " 'eutropius/eutropius10.txt',\n",
       " 'eutropius/eutropius2.txt',\n",
       " 'eutropius/eutropius3.txt',\n",
       " 'eutropius/eutropius4.txt',\n",
       " 'eutropius/eutropius5.txt',\n",
       " 'eutropius/eutropius6.txt',\n",
       " 'eutropius/eutropius7.txt',\n",
       " 'eutropius/eutropius8.txt',\n",
       " 'eutropius/eutropius9.txt',\n",
       " 'prudentius/prud.psycho.txt',\n",
       " 'prudentius/prud1.txt',\n",
       " 'prudentius/prud10.txt',\n",
       " 'prudentius/prud11.txt',\n",
       " 'prudentius/prud12.txt',\n",
       " 'prudentius/prud13.txt',\n",
       " 'prudentius/prud14.txt',\n",
       " 'prudentius/prud2.txt',\n",
       " 'prudentius/prud3.txt',\n",
       " 'prudentius/prud4.txt',\n",
       " 'prudentius/prud5.txt',\n",
       " 'prudentius/prud6.txt',\n",
       " 'prudentius/prud7.txt',\n",
       " 'prudentius/prud8.txt',\n",
       " 'prudentius/prud9.txt',\n",
       " 'suetonius/suet.caesar.txt',\n",
       " 'xylander/caesar.txt']"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "latin_reader = get_corpus_reader('latin_text_latin_library', language='latin')\n",
    "ALL_FILE_IDS = list(latin_reader.fileids())\n",
    "good_files = [file for file in ALL_FILE_IDS\n",
    "              if 'prudentius' in file or\n",
    "              'caesar' in file or\n",
    "              'eutropius' in file]\n",
    "LOG.info('available good files %s', len(good_files))\n",
    "latin_reader._fileids = good_files\n",
    "good_files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# remove some unfamiliar entries\n",
    "questionable = ['caesar/alex.txt',\n",
    "                'caesar/hisp.txt',\n",
    "                'prudentius/prud.psycho.txt',\n",
    "                'suetonius/suet.caesar.txt',\n",
    "                'xylander/caesar.txt']\n",
    "for file in questionable:\n",
    "    good_files.remove(file)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define a custom Scikit-learn Pipeline, and call the CorpusReader `words()` method to process the texts\n",
    "#### The functions used in the pipelines are doctest documented in the `corpus_cleaning` module\n",
    "#### The functions used and their order was developed iteratively by running the pipelines on actual data and carefully inspecting the results prior to runnning it through featurization. Always know your data!\n",
    "\n",
    "#### Lastly, we use the joblib library to save/pickle the pipeline so that it can be reloaded and reused."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['process_latin_text_pipeline.0.22.1.joblib']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "process_latin_text_pipeline = Pipeline([\n",
    "    ('separate_camel_cases', FunctionTransformer(separate_camel_cases, validate=False)),\n",
    "    ('splice_hyphens', FunctionTransformer(splice_hyphens, validate=False)),\n",
    "    ('jv_transform', FunctionTransformer(jv_transform, validate=False)),\n",
    "    ('aeoe_transform', FunctionTransformer(aeoe_transform, validate=False)),\n",
    "    ('accept_editorial', FunctionTransformer(accept_editorial, validate=False)),\n",
    "    ('drop_enclitics', FunctionTransformer(drop_enclitics, validate=False)),\n",
    "    ('drop_fringe_punctuation', FunctionTransformer(drop_fringe_punctuation, validate=False)),\n",
    "    ('drop_all_punctuation', FunctionTransformer(drop_all_punctuation, validate=False)),\n",
    "    ('drop_non_lower', FunctionTransformer(drop_non_lower, validate=False)),\n",
    "    ('drop_arabic_numeric', FunctionTransformer(drop_arabic_numeric, validate=False)),\n",
    "    ('drop_all_caps', FunctionTransformer(drop_all_caps, validate=False)),\n",
    "    ('divide_separate_words', FunctionTransformer(divide_separate_words, validate=False)),\n",
    "    ('drop_empty_lists', FunctionTransformer(drop_empty_lists, validate=False)),\n",
    "    ('drop_empty_strings', FunctionTransformer(drop_empty_strings, validate=False))])\n",
    "\n",
    "process_latin_text_pipeline_file = 'process_latin_text_pipeline.{}.joblib'.format(\n",
    "    sklearn.__version__)\n",
    "dump(process_latin_text_pipeline, process_latin_text_pipeline_file)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00, 27.98it/s]\n"
     ]
    }
   ],
   "source": [
    "X = process_latin_text_pipeline.fit_transform(tqdm([list(latin_reader.words())]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Analyze the resulting matrix, by profiling the character occurences\n",
    "* Go back and adjust the pipeline as necessary\n",
    "* Turn the output into a distinct set of words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Character distribution profile, total chars: 810409\n",
      "Counter({'e': 93828, 'i': 90246, 'u': 75553, 't': 67540, 'a': 65821, 's': 62825, 'r': 54357, 'n': 49472, 'o': 44779, 'm': 41221, 'c': 31563, 'p': 23317, 'l': 23269, 'd': 20970, 'b': 13064, 'q': 12561, 'g': 8357, 'f': 7372, 'x': 4467, 'h': 4172, 'C': 2515, 'A': 1473, 'I': 1185, 'P': 1119, 'H': 1079, 'S': 1078, 'R': 965, 'G': 770, 'Q': 607, 'M': 601, 'T': 557, 'U': 546, 'N': 529, 'L': 475, 'E': 469, 'D': 459, 'y': 399, 'B': 351, 'F': 211, 'O': 203, 'z': 23, 'Z': 16, 'K': 11, 'k': 10, 'X': 3, 'Å': 1})\n",
      "Number of distinct words in Eutropius/Prudentius/Caesar sample: 25,773\n",
      "Number of distinct characters: 46\n"
     ]
    }
   ],
   "source": [
    "char_count = profile_chars(X)\n",
    "print('Character distribution profile, total chars:', sum(char_count.values()))\n",
    "print(char_count)\n",
    "distinct_good_latin = distinct_words(X)\n",
    "print(f'Number of distinct words in Eutropius/Prudentius/Caesar sample: {len(distinct_good_latin):,}')\n",
    "print(f'Number of distinct characters: {len(char_count)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### After running this notebook several times, we've decided to load in more training data, which is provide by the notebook:\n",
    "* `boosting_training_data.ipynb`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['recusabunt', 'duritiis', 'adfluentibus', 'Pyramus', 'linteolis']"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "additional_latin_words = []\n",
    "with open('latin.lemma.forms.txt', 'rt') as reader:\n",
    "    additional_latin_words = reader.read().split('\\n')\n",
    "random.sample(additional_latin_words, 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "additional_latin_words: 175,970\n",
      "distinct_good_latin now: 179,919\n"
     ]
    }
   ],
   "source": [
    "print(f'additional_latin_words: {len(additional_latin_words):,}')\n",
    "distinct_good_latin= list(set(distinct_good_latin) | set(additional_latin_words))\n",
    "print(f'distinct_good_latin now: {len(distinct_good_latin):,}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Load the Greek texts of Homer and Plato (two of the most commonly quoted Greek authors)\n",
    "* Preprocess the text\n",
    "* Transliterate into Classical Latin\n",
    "\n",
    "#### We save this pipeline for reuse too."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['process_greek_text_pipeline.0.22.1.joblib']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "perseus_greek = get_corpus_reader(language='greek', corpus_name='greek_text_perseus')\n",
    "plato = [tmp for tmp in perseus_greek.fileids() if 'plato' in tmp]\n",
    "homer = [tmp for tmp in perseus_greek.fileids() if 'homer' in tmp]\n",
    "greek_texts = plato + homer\n",
    "\n",
    "process_greek_pipeline = Pipeline([\n",
    "    ('accept_editorial', FunctionTransformer(accept_editorial, validate=False)),  # problematic\n",
    "    ('romanizer', FunctionTransformer(romanizer_transform, validate=False)),\n",
    "    ('drop_fringe_punctuation', FunctionTransformer(drop_fringe_punctuation, validate=False)),\n",
    "    ('drop_all_punctuation', FunctionTransformer(drop_all_punctuation, validate=False)),\n",
    "    ('drop_arabic_numeric', FunctionTransformer(drop_arabic_numeric, validate=False)),  #ok\n",
    "    ('drop_empty_lists', FunctionTransformer(drop_empty_lists, validate=False)),  # problem?\n",
    "    ('drop_empty_strings', FunctionTransformer(drop_empty_strings, validate=False))  # problem?\n",
    "])\n",
    "\n",
    "process_greek_text_pipeline_file = 'process_greek_text_pipeline.{}.joblib'.format(\n",
    "    sklearn.__version__)\n",
    "dump(process_greek_pipeline, process_greek_text_pipeline_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00,  9.64it/s]\n"
     ]
    }
   ],
   "source": [
    "X_greek_transliterated = process_greek_pipeline.fit_transform(tqdm([list(perseus_greek.words(greek_texts))]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Analyze the transliterated Greek examples\n",
    "* Check character profiles for tuning\n",
    "* Create a set distinct words, with and without macrons"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Character distribution profile of transliterated Greek:  Counter({'a': 176222, 'e': 172410, 'o': 159911, 'i': 158418, 'n': 143139, 't': 142875, 's': 107062, 'h': 91529, 'u': 74411, 'p': 66229, 'k': 63931, 'ē': 63161, 'r': 56352, 'm': 52545, 'l': 49342, 'ō': 48812, 'd': 42019, 'g': 29246, 'b': 6357, 'x': 4571, 'S': 4393, 'E': 3738, 'Ō': 3736, 'A': 3441, 'z': 2850, 'T': 2102, 'P': 1859, 'I': 1327, 'R': 1056, 'X': 1044, 'K': 1004, 'L': 979, 'O': 859, 'D': 743, 'N': 667, 'M': 651, 'H': 461, 'Z': 234, 'Ē': 164, 'Y': 143, 'G': 53, 'B': 35, 'y': 23, 'F': 11, 'f': 5, 'c': 5, ' ': 4, 'V': 2, 'C': 2, 'v': 2, 'U': 1})\n",
      "41,552 distinct_transliterated_greek_examples\n",
      "39,982 distinct_demacronized_greek\n"
     ]
    }
   ],
   "source": [
    "greek_char_count = profile_chars(X_greek_transliterated)\n",
    "print('Character distribution profile of transliterated Greek: ', greek_char_count )\n",
    "distinct_transliterated_greek_examples = distinct_words(X_greek_transliterated)\n",
    "print(f'{len(distinct_transliterated_greek_examples):,} distinct_transliterated_greek_examples')\n",
    "distinct_demacronized_greek = distinct_words(demacronize(X_greek_transliterated))\n",
    "print(f'{len(distinct_demacronized_greek):,} distinct_demacronized_greek')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total_uniq_char_count 54\n"
     ]
    }
   ],
   "source": [
    "char_idx = {' ':0}\n",
    "curr_idx=1\n",
    "\n",
    "all_chars = sorted(list(char_count.keys()) + list(greek_char_count.keys()))\n",
    "\n",
    "for car in all_chars:\n",
    "    if car not in char_idx:\n",
    "        char_idx[car] =curr_idx\n",
    "        curr_idx+=1\n",
    "        \n",
    "\n",
    "total_uniq_char_count = len(char_idx )\n",
    "print('total_uniq_char_count', total_uniq_char_count)\n",
    "\n",
    "\n",
    "def encode_word(word, char_idx, max_len=28):\n",
    "    vals =np.zeros(max_len)\n",
    "    for idx, car in enumerate(word):\n",
    "        vals[idx] = char_idx.get(car, 0) / len(char_idx) # total_uniq_char_count\n",
    "    return vals\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "54"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(char_idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{' ': 0, 'A': 1, 'B': 2, 'C': 3, 'D': 4, 'E': 5, 'F': 6, 'G': 7, 'H': 8, 'I': 9, 'K': 10, 'L': 11, 'M': 12, 'N': 13, 'O': 14, 'P': 15, 'Q': 16, 'R': 17, 'S': 18, 'T': 19, 'U': 20, 'V': 21, 'X': 22, 'Y': 23, 'Z': 24, 'a': 25, 'b': 26, 'c': 27, 'd': 28, 'e': 29, 'f': 30, 'g': 31, 'h': 32, 'i': 33, 'k': 34, 'l': 35, 'm': 36, 'n': 37, 'o': 38, 'p': 39, 'q': 40, 'r': 41, 's': 42, 't': 43, 'u': 44, 'v': 45, 'x': 46, 'y': 47, 'z': 48, 'Å': 49, 'Ē': 50, 'ē': 51, 'Ō': 52, 'ō': 53}\n"
     ]
    }
   ],
   "source": [
    "print( char_idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "the_char_idx = {' ': 0, 'A': 1, 'B': 2, 'C': 3, 'D': 4, 'E': 5, 'F': 6, 'G': 7, 'H': 8, 'I': 9, 'K': 10, 'L': 11, 'M': 12, 'N': 13, 'O': 14, 'P': 15, 'Q': 16, 'R': 17, 'S': 18, 'T': 19, 'U': 20, 'V': 21, 'X': 22, 'Y': 23, 'Z': 24, 'a': 25, 'b': 26, 'c': 27, 'd': 28, 'e': 29, 'f': 30, 'g': 31, 'h': 32, 'i': 33, 'k': 34, 'l': 35, 'm': 36, 'n': 37, 'o': 38, 'p': 39, 'q': 40, 'r': 41, 's': 42, 't': 43, 'u': 44, 'v': 45, 'x': 46, 'y': 47, 'z': 48, 'Å': 49, 'Ē': 50, 'ē': 51, 'Ō': 52, 'ō': 53}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### See how many words from the transliterated Greek words which have also appear in the Latin corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shared_words: 524 : {'anathema', 'demo', 'Rhodope', 'damna', 'Antiphates', 'Euenos', 'dogma', 'toros', 'patres', 'Atrei', 'e', 'pedo', 'philo', 'phono', 'nome', 'Dardanide', 'parentes', 'in', 'dote', 'neo', 'xanthe', 'metis', 'agre', 'eges', 'muri', 'sito', 'peribolo', 'oro', 'Diomedeos', 'nassa', 'di', 'r', 'eri', 'notio', 'pales', 'hamos', 'mel', 'throno', 'pleto', 'Herme', 'agrio', 'ara', 'esto', 'Eos', 'ore', 'Argos', 'philosopho', 'audes', 'emat', 'Atreus', 'spei', 'pio', 'med', 'doma', 'melle', 'esset', 'ose', 'Amphitrites', 'asto', 'sus', 'det', 'geode', 'exite', 'eris', 'plana', 'aloes', 'Parmenide', 'tris', 'ergo', 'trite', 'tauros', 'manenti', 'paries', 'Thesea', 'ale', 'Delo', 'lego', 'nota', 'sto', 'poma', 'dogmata', 'aloe', 'erga', 'daphne', 'elusa', 'os', 'osi', 'eras', 'Lampetie', 'ites', 'Orion', 'io', 'Lapithas', 'bolos', 'mela', 'ae', 'polles', 'times', 'erato', 'lino', 'heroi', 'agro', 'hei', 'noma', 'Thebe', 'mori', 'Apollo', 'hesperos', 'philosophos', 'problemata', 'deo', 'aleis', 'elateri', 'Xanthippe', 'beta', 'meri', 'non', 'leges', 'sin', 'ante', 'bie', 'potero', 'horas', 'Agamemnonides', 'Die', 'metreta', 'Sparte', 'Hesiodo', 'agoni', 'an', 'arere', 'etis', 'Stesichorus', 'Minoa', 'dromo', 'agnos', 'prores', 'Polites', 'molis', 'Lede', 'Amphitrite', 'at', 'saturo', 'par', 'melie', 'eget', 'Euripide', 'parergo', 'hora', 'ted', 'mi', 'lexeos', 'sei', 'Same', 'ipsa', 'Mida', 'Athenas', 'aristas', 'dei', 'tria', 'mete', 'Parthenio', 'apate', 'temo', 'athletas', 'domo', 'ponto', 'alto', 'horai', 'Phalereus', 'a', 'pauet', 'pedalia', 'exerat', 'alete', 'Selene', 'lethe', 'meis', 'Homero', 'anterotas', 'Delio', 'hos', 'exe', 'polite', 'strato', 'polo', 'nae', 'deis', 'ea', 'per', 'ereto', 'lupe', 'duas', 'suos', 'lis', 'Troia', 'Europes', 'pono', 'geometre', 'aget', 'Hippotades', 'lae', 'mala', 'lita', 'hiet', 'has', 'Persephone', 'patris', 'Protagoras', 'Maia', 'do', 'me', 'hero', 'Aretes', 'aedes', 'ambrosie', 'proi', 'thalamo', 'erosa', 'Theodore', 'Troes', 'orsas', 'leget', 'Leontinos', 'heroas', 'heroes', 'Elpenor', 'Ossa', 'Dione', 'Hos', 'oleto', 'theatro', 'Thebes', 'id', 'et', 'lebes', 'metri', 'patria', 'Laertes', 'Pheres', 'Hermes', 'morio', 'ese', 'meti', 'hippo', 'hemeris', 'eant', 'en', 'alueis', 'dein', 'lege', 'phthisis', 'Minos', 'periodo', 'thalle', 'Triptolemo', 'rhetor', 'prophetis', 'axo', 'penes', 'Megapenthes', 'mero', 'mese', 'Euripides', 'mede', 'ito', 'logo', 'Thebas', 'pater', 'Hermionen', 'anathemata', 'osse', 'limen', 'ant', 'Epigenes', 'dele', 'geme', 'siderea', 'dogmati', 'mones', 'Diones', 'Tantalo', 'historia', 'opi', 'Eroti', 'aletes', 'thalamos', 'elato', 'ero', 'Eo', 'die', 'Argo', 'labe', 'patri', 'dis', 'Eros', 'mere', 'eis', 'enen', 'Nestor', 'antas', 'epistate', 'sape', 'tende', 'Letoi', 'messe', 'metalla', 'aure', 'dolo', 'iambe', 'Asia', 'esse', 'duo', 'o', 'phantasia', 'opta', 'Laertiade', 'labes', 'nu', 'edos', 'stes', 'protero', 'Ares', 'Perseus', 'domat', 'Anaxagoras', 'nomisma', 'pedalio', 'philosophia', 'te', 'Orpheus', 'emata', 'laro', 'Cypria', 'Apolloni', 'rhetores', 'psephismata', 'ere', 'Gorgias', 'liga', 'Neleus', 'aniso', 'eita', 'auge', 'heroos', 'lu', 'emes', 'de', 'homo', 'hostis', 'ne', 'orse', 'pie', 'Are', 'eo', 'Mimas', 'Zenoni', 'Aristophane', 'organo', 'prophetas', 'is', 'Paros', 'ares', 'bio', 'Naxo', 'Naxos', 'diae', 'phantasma', 'depote', 'es', 'phantasmata', 'Plato', 'Eoi', 'Lesbo', 'pedali', 'elephanti', 'aut', 'pro', 'Menelao', 'aeto', 'assa', 'dies', 'Hermogene', 'astra', 'potho', 'pale', 'metro', 'stephano', 'probata', 'Erote', 'hebe', 'diastemata', 'dine', 'pompas', 'Miletos', 'telamon', 'Gorgia', 'Delon', 'polle', 'psoras', 'oles', 'dolos', 'polos', 'eas', 'domos', 'emo', 'Pausania', 'olet', 'Hellesponto', 'Melete', 'tot', 'Europe', 'Admete', 'elate', 'topazo', 'no', 'isto', 'luto', 'statera', 'mone', 'Arete', 'tropo', 'Leto', 'topo', 'auges', 'pares', 'Aristodemo', 'ales', 'Lemnos', 'metabole', 'osa', 'elephantos', 'ego', 'hebes', 'hoste', 'limo', 'Theodoro', 'age', 'axioma', 'noto', 'admete', 'Gai', 'ede', 'Helene', 'aetas', 'omen', 'glossas', 'iota', 'rhetori', 'daemones', 'heros', 'se', 'ago', 'bibas', 'doto', 'dido', 'ambrosia', 'torno', 'aera', 'sidereos', 'sues', 'petas', 'Theseus', 'adamas', 'poesi', 'eos', 'Pergama', 'zeta', 'Euripo', 'idea', 'dos', 'Protagora', 'melo', 'agones', 'asbestos', 'peren', 'lotos', 'aer', 'daphnes', 'Eous', 'genus', 'aude', 'meni', 'aule', 'heroa', 'suas', 'Neritos', 'est', 'deos', 'orto', 'pege', 'Delos', 'Perses', 'nautas', 'problema', 'Orestes', 'Aristotele', 'agonia', 'porro', 'geometria', 'iste', 'pare', 'ei', 'antro', 'aspis', 'seu', 'ex', 'pino', 'Phemio', 'stantes', 'erat', 'Pelion', 'pharetras', 'halo', 'eripe', 'diametro', 'ano', 'ambrosio', 'mera', 'Melite', 'Melantho', 'pote', 'pe', 'Troas'}\n"
     ]
    }
   ],
   "source": [
    "shared_words = distinct_demacronized_greek & set(distinct_good_latin)\n",
    "print(f'Shared_words: {len(shared_words)} : {shared_words}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### These shared words appear in both language corpora; however, intuitively, we know each word will have a different probability of occurrence in each language. So, rather than arbitrarily excluding some or all of the words from one language or the other, we should split them into the most common probable groups. We can do this by loading the probability distribution pickle objects we have created in the notebooks:\n",
    "* `building_language_model/make_frequency_distribution.ipynb` \n",
    "* `detecting_loanwords/make_frequency_distribution_greek_transliterated.ipynb`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load Frequency Distributions for Latin and transliterated Greek"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "greek_transliterated_word_probs = {}\n",
    "with open('freq_dist.greek.transliterated.pkl', 'rb') as reader:\n",
    "    greek_transliterated_word_probs = pickle.load(reader)\n",
    "    \n",
    "latin_word_probs = {}\n",
    "with open(os.path.join('../probablistic_language_modeling', 'freq_dist.latin.pkl'), 'rb') as reader:\n",
    "    latin_word_probs = pickle.load(reader)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### We'll create a list of tuples containing (the word, the words probability in Latin, the words probability in Greek)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('et', 0.9999900000000002, 0.001594731640385874)\n",
      "('in', 0.6195366094371269, 0.002322993118442995)\n",
      "('est', 0.40047688283934835, 0.0015525135836869104)\n",
      "('non', 0.3655202190736462, 1.0000000000000006e-06)\n",
      "('de', 0.16358695886651264, 0.5190825616279315)\n",
      "('a', 0.13448064535122392, 0.002914045912228485)\n",
      "('ex', 0.13319985495959372, 0.02580678715724147)\n",
      "('per', 0.11551978686121114, 0.0049721761763029575)\n",
      "('esse', 0.11489112051513624, 1e-06)\n",
      "('se', 0.10337573584020457, 0.019231324826377898)\n",
      "('aut', 0.09289014417246275, 0.0011619965592214974)\n",
      "('me', 0.06858796751076134, 0.016392160513372597)\n",
      "('te', 0.068280671647568, 0.1623927550925633)\n",
      "('id', 0.059277606586987885, 0.00039151702446541285)\n",
      "('ne', 0.05901722612275537, 2.2109028349481774e-05)\n",
      "('eo', 0.05369232834070303, 1.1554514174740887e-05)\n",
      "('pro', 0.051006421750196464, 0.00823352105629789)\n",
      "('ea', 0.04345773405740157, 0.0006131618221349714)\n",
      "('erat', 0.03941362666697944, 1e-06)\n",
      "('ei', 0.034644676362612745, 0.07388259922318621)\n",
      "('esset', 0.03463529328282058, 5.3772570873704435e-05)\n",
      "('ego', 0.0335984629657866, 1e-06)\n",
      "('eos', 0.03038241236702284, 1e-06)\n",
      "('ante', 0.030340188507958103, 0.00012765417009689064)\n",
      "('ergo', 0.029326815890404534, 1e-06)\n",
      "('an', 0.027738729635580994, 0.10501841603867183)\n",
      "('ipsa', 0.021074397213197437, 1.1554514174740887e-05)\n",
      "('die', 0.019035923128350087, 8.54361133979271e-05)\n",
      "('eis', 0.018604301457910602, 0.08208345673695988)\n",
      "('dies', 0.015604061694366579, 1e-06)\n",
      "('is', 0.015097375385589792, 0.0012464326726194246)\n",
      "('e', 0.014745509893383692, 0.0011619965592214974)\n",
      "('pater', 0.01401128389964696, 0.0017002767821332827)\n",
      "('homo', 0.012092444082149687, 2.2109028349481774e-05)\n",
      "('suos', 0.010846840239740087, 9.5990627572668e-05)\n",
      "('es', 0.010710785582753728, 0.1048812073544002)\n",
      "('at', 0.010438676268781009, 0.0005287257087370443)\n",
      "('duo', 0.009596544857434407, 0.010365532919595551)\n",
      "('genus', 0.009535554838785348, 1.1554514174740887e-05)\n",
      "('patris', 0.00863712494868577, 0.00028597188271800393)\n",
      "('seu', 0.008449463352842516, 0.00016987222679585418)\n",
      "('lege', 0.00787240394562451, 0.0016580587254343192)\n",
      "('tot', 0.00781141392697545, 0.002069684778249214)\n",
      "('dei', 0.007619060791236115, 0.012793071179785955)\n",
      "('eas', 0.007398558416120292, 0.00010654514174740887)\n",
      "('o', 0.00659395932394234, 0.0006131618221349714)\n",
      "('suas', 0.006291355000645093, 5.3772570873704435e-05)\n",
      "('deos', 0.006263205761268605, 0.001109223988347793)\n",
      "('ore', 0.005735407522959452, 1e-06)\n",
      "('mala', 0.005711949823479046, 0.00654479878833935)\n",
      "('deo', 0.005665034424518232, 1e-06)\n",
      "('iste', 0.005374158950961189, 0.001014233360775125)\n",
      "('meis', 0.0050715546276639415, 2.2109028349481774e-05)\n",
      "('hos', 0.004811174163431426, 0.022239361366179048)\n",
      "('domo', 0.004691539896081352, 1e-06)\n",
      "('leges', 0.0043537490235634945, 1e-06)\n",
      "('mi', 0.004102751639123142, 1.1554514174740887e-05)\n",
      "('patres', 0.004001883531357393, 1e-06)\n",
      "('patri', 0.003940893512708336, 0.0019958031790260275)\n",
      "('duas', 0.00381656770546218, 6.432708504844532e-05)\n",
      "('dolo', 0.0035163091521129737, 1e-06)\n",
      "('has', 0.0033943291148148583, 0.0036211983619361243)\n",
      "('hostis', 0.003326301786321679, 0.004433895953391172)\n",
      "('isto', 0.00287825972624591, 1e-06)\n",
      "('porro', 0.0028243070174409743, 1e-06)\n",
      "('par', 0.002765662768739957, 0.016128297659004077)\n",
      "('tria', 0.002758625458895835, 0.0014680774702889832)\n",
      "('patria', 0.0027492423791036728, 0.0003176354252422266)\n",
      "('sin', 0.002629608111753598, 3.266354252422266e-05)\n",
      "('mori', 0.0026084961822212317, 1.0000000000000006e-06)\n",
      "('aetas', 0.0025662723231565, 1e-06)\n",
      "('erga', 0.002556889243364337, 0.00464498623688599)\n",
      "('os', 0.002521702694143727, 0.00010654514174740887)\n",
      "('hora', 0.0024912076848191984, 0.001584177126211133)\n",
      "('di', 0.0024137972765338562, 0.014207376079201233)\n",
      "('nota', 0.0024067599666897335, 1e-06)\n",
      "('parentes', 0.0021862575915739106, 2.2109028349481774e-05)\n",
      "('agro', 0.0021088471832885675, 1e-06)\n",
      "('esto', 0.001998595995730656, 1e-06)\n",
      "('hoste', 0.0018461209491080122, 5.3772570873704435e-05)\n",
      "('domos', 0.0018132801698354428, 0.00012765417009689064)\n",
      "('penes', 0.0017053747522255717, 1e-06)\n",
      "('eris', 0.0016021608745117818, 0.00018042674097059507)\n",
      "('dein', 0.001588086254823538, 0.0036211983619361243)\n",
      "('alto', 0.001550553935654887, 5.3772570873704435e-05)\n",
      "('dis', 0.001522404696278399, 0.001404750385240538)\n",
      "('aedes', 0.0014801808372136667, 1e-06)\n",
      "('en', 0.0014403027480969753, 0.1989535921938657)\n",
      "('ero', 0.0014004246589802837, 1e-06)\n",
      "('Eo', 0.0013417804102792668, 1e-06)\n",
      "('Asia', 0.0013159769408508193, 0.0007925885631055665)\n",
      "('pares', 0.0012479496123576398, 1.1554514174740887e-05)\n",
      "('age', 0.0012479496123576398, 0.0016052861545606148)\n",
      "('ant', 0.0011728849740203382, 0.001014233360775125)\n",
      "('det', 0.0011588103543320942, 1e-06)\n",
      "('aer', 0.0011588103543320942, 1e-06)\n",
      "('aera', 0.001130661114955606, 0.0003070809110674857)\n",
      "('astra', 0.0011025118755791177, 0.00016987222679585418)\n",
      "('dote', 0.001013372617553572, 0.00021209028349481775)\n",
      "('damna', 0.0009617656786966772, 1e-06)\n",
      "('spei', 0.0009570741388005958, 1e-06)\n",
      "('Plato', 0.000912504509787823, 2.2109028349481774e-05)\n",
      "('oro', 0.0008890468103074162, 1e-06)\n",
      "('philosophia', 0.0008843552704113349, 0.0003492989677664493)\n",
      "('potero', 0.0008820095004632942, 1e-06)\n",
      "('Hos', 0.0008608975709309282, 9.5990627572668e-05)\n",
      "('historia', 0.0008421314113466028, 0.0001487631984463724)\n",
      "('daemones', 0.0008280567916583586, 1e-06)\n",
      "('Apollo', 0.0008045990921779519, 1e-06)\n",
      "('dos', 0.0007717583129053824, 0.0005181711945623034)\n",
      "('eras', 0.0007576836932171383, 0.00011709965592214976)\n",
      "('ago', 0.000745954843476935, 1.0000000000000006e-06)\n",
      "('muri', 0.0007107682942563249, 6.432708504844532e-05)\n",
      "('idea', 0.0006615071253474707, 0.0002437538260190404)\n",
      "('horas', 0.0006450867357111859, 0.0016475042112595783)\n",
      "('do', 0.0006192832662827385, 0.0001487631984463724)\n",
      "('deis', 0.0005958255668023318, 1.1554514174740887e-05)\n",
      "('r', 0.0005747136372699656, 0.0001593177126211133)\n",
      "('poma', 0.0005606390175817216, 1.1554514174740887e-05)\n",
      "('pio', 0.0005536017077375996, 1e-06)\n",
      "('lis', 0.000551255937789559, 2.2109028349481774e-05)\n",
      "('med', 0.0005418728579973961, 1e-06)\n",
      "('omen', 0.0005348355481532742, 1e-06)\n",
      "('limen', 0.0005277982383091521, 2.2109028349481774e-05)\n",
      "('lego', 0.0005207609284650301, 1e-06)\n",
      "('Athenas', 0.0005184151585169894, 1e-06)\n",
      "('mero', 0.0005184151585169894, 1e-06)\n",
      "('melle', 0.0005160693885689487, 1.0000000000000006e-06)\n",
      "('prophetas', 0.0005019947688807047, 1e-06)\n",
      "('tris', 0.0004902659191405013, 0.0007503705064066029)\n",
      "('ara', 0.0004879201491924607, 0.017859237983661583)\n",
      "('ponto', 0.0004644624497120539, 1e-06)\n",
      "('Troia', 0.0004550793699198912, 0.00020153576932007686)\n",
      "('probata', 0.0004480420600757691, 0.000655379878833935)\n",
      "('luto', 0.00043631321033556577, 9.5990627572668e-05)\n",
      "('heros', 0.0004292759004914438, 1e-06)\n",
      "('mel', 0.0004245843605953624, 1e-06)\n",
      "('philosophos', 0.0004245843605953624, 0.0002648628543685222)\n",
      "('prophetis', 0.0004222385906473217, 1e-06)\n",
      "('theatro', 0.0004198928206992811, 1e-06)\n",
      "('times', 0.000412855510855159, 1.0000000000000006e-06)\n",
      "('ales', 0.000412855510855159, 1.0000000000000006e-06)\n",
      "('plana', 0.0004081639709590777, 0.00010654514174740887)\n",
      "('Perseus', 0.000405818201011037, 0.00048650765203808076)\n",
      "('Eos', 0.0004034724310629963, 1e-06)\n",
      "('aure', 0.00040112666111495566, 1e-06)\n",
      "('opi', 0.0003847062714786709, 0.00012765417009689064)\n",
      "('pie', 0.0003847062714786709, 1.1554514174740887e-05)\n",
      "('antro', 0.00038001473158258956, 1e-06)\n",
      "('Die', 0.0003753231916865082, 1e-06)\n",
      "('dolos', 0.00037297742173846753, 7.488159922318621e-05)\n",
      "('eget', 0.0003659401118943455, 1e-06)\n",
      "('audes', 0.00035421126215414215, 1e-06)\n",
      "('Thebas', 0.0003448281823619794, 1e-06)\n",
      "('polo', 0.000340136642465898, 1e-06)\n",
      "('thalamos', 0.00033075356267373533, 6.432708504844532e-05)\n",
      "('Argos', 0.000326062022777654, 0.0010775604458235703)\n",
      "('limo', 0.0003143331730374506, 1e-06)\n",
      "('stantes', 0.0003072958631933286, 1.1554514174740887e-05)\n",
      "('aget', 0.00029087547355704385, 9.5990627572668e-05)\n",
      "('philosopho', 0.0002791466238168405, 1e-06)\n",
      "('tauros', 0.0002744550839207592, 0.00020153576932007686)\n",
      "('elephanti', 0.0002744550839207592, 0.00010654514174740887)\n",
      "('orto', 0.00027210931397271846, 1e-06)\n",
      "('elephantos', 0.00026976354402467775, 0.0005287257087370443)\n",
      "('paries', 0.0002580346942844744, 1e-06)\n",
      "('toros', 0.00024630584454427105, 1.1554514174740887e-05)\n",
      "('agnos', 0.0002439600745962303, 1.0000000000000006e-06)\n",
      "('labe', 0.0002439600745962303, 0.0003492989677664493)\n",
      "('Homero', 0.00023926853470014893, 1e-06)\n",
      "('ae', 0.00023223122485602693, 3.266354252422266e-05)\n",
      "('metalla', 0.0002251939150119049, 0.00010654514174740887)\n",
      "('thalamo', 0.00022050237511582355, 1e-06)\n",
      "('io', 0.00021815660516778286, 1.0000000000000006e-06)\n",
      "('molis', 0.00021346506527170154, 0.0017108312963080236)\n",
      "('Theseus', 0.0002087735253756202, 1e-06)\n",
      "('pote', 0.00020408198547953882, 0.012581980896291136)\n",
      "('Aristotele', 0.00019939044558345747, 1e-06)\n",
      "('mera', 0.0001970446756354168, 1e-06)\n",
      "('nautas', 0.00018297005594717272, 0.0001487631984463724)\n",
      "('anathema', 0.00017827851605109138, 3.266354252422266e-05)\n",
      "('dogma', 0.00017358697615501003, 0.0003176354252422266)\n",
      "('Minos', 0.00017358697615501003, 1e-06)\n",
      "('eripe', 0.00017358697615501003, 1e-06)\n",
      "('dogmata', 0.00017124120620696934, 4.321805669896355e-05)\n",
      "('Orpheus', 0.00017124120620696934, 0.0001593177126211133)\n",
      "('notio', 0.00016889543625892866, 1e-06)\n",
      "('throno', 0.00016889543625892866, 1e-06)\n",
      "('sus', 0.00016889543625892866, 0.00012765417009689064)\n",
      "('labes', 0.00016889543625892866, 1e-06)\n",
      "('sues', 0.00016889543625892866, 0.00010654514174740887)\n",
      "('Pergama', 0.00016889543625892866, 1.0000000000000006e-06)\n",
      "('ere', 0.0001642038963628473, 1.0000000000000006e-06)\n",
      "('axioma', 0.0001642038963628473, 1e-06)\n",
      "('phantasma', 0.00016185812641480665, 8.54361133979271e-05)\n",
      "('petas', 0.00016185812641480665, 1.0000000000000006e-06)\n",
      "('Orestes', 0.00016185812641480665, 1e-06)\n",
      "('eant', 0.0001571665865187253, 1.1554514174740887e-05)\n",
      "('no', 0.0001571665865187253, 0.0003598534819411902)\n",
      "('noto', 0.00015482081657068462, 1e-06)\n",
      "('pauet', 0.00015012927667460324, 1.0000000000000006e-06)\n",
      "('hebes', 0.00015012927667460324, 1e-06)\n",
      "('emat', 0.0001454377367785219, 1e-06)\n",
      "('Orion', 0.0001384004269343999, 1e-06)\n",
      "('Argo', 0.0001384004269343999, 1e-06)\n",
      "('rhetor', 0.00013605465698635923, 1e-06)\n",
      "('leget', 0.00013370888703831852, 1.0000000000000006e-06)\n",
      "('metro', 0.00013370888703831852, 1e-06)\n",
      "('Thesea', 0.00012901734714223717, 1e-06)\n",
      "('pono', 0.00012901734714223717, 1.0000000000000006e-06)\n",
      "('Ossa', 0.0001266715771941965, 1e-06)\n",
      "('olet', 0.0001266715771941965, 1.0000000000000006e-06)\n",
      "('lino', 0.00012432580724615583, 1e-06)\n",
      "('mones', 0.00012198003729811518, 1e-06)\n",
      "('domat', 0.00012198003729811518, 1e-06)\n",
      "('Hellesponto', 0.00011728849740203382, 1e-06)\n",
      "('Delo', 0.00011494272745399313, 1e-06)\n",
      "('Gai', 0.00011259695750595246, 3.266354252422266e-05)\n",
      "('geometria', 0.00011259695750595246, 1e-06)\n",
      "('pharetras', 0.00011259695750595246, 2.2109028349481774e-05)\n",
      "('Perses', 0.00011025118755791177, 1e-06)\n",
      "('ted', 0.0001079054176098711, 1e-06)\n",
      "('strato', 0.0001079054176098711, 1e-06)\n",
      "('Rhodope', 0.00010555964766183044, 1e-06)\n",
      "('ede', 0.00010555964766183044, 1e-06)\n",
      "('aristas', 9.852233781770841e-05, 6.432708504844532e-05)\n",
      "('Dione', 9.852233781770841e-05, 1e-06)\n",
      "('eri', 9.383079792162705e-05, 1e-06)\n",
      "('mela', 9.148502797358637e-05, 1e-06)\n",
      "('meri', 9.148502797358637e-05, 1.0000000000000006e-06)\n",
      "('Euripides', 9.148502797358637e-05, 1.1554514174740887e-05)\n",
      "('Anaxagoras', 8.91392580255457e-05, 5.3772570873704435e-05)\n",
      "('polos', 8.91392580255457e-05, 1e-06)\n",
      "('Pelion', 8.91392580255457e-05, 1e-06)\n",
      "('proi', 8.679348807750503e-05, 1e-06)\n",
      "('bibas', 8.679348807750503e-05, 3.266354252422266e-05)\n",
      "('Maia', 8.444771812946434e-05, 7.488159922318621e-05)\n",
      "('rhetores', 8.444771812946434e-05, 1e-06)\n",
      "('phantasmata', 8.444771812946434e-05, 4.321805669896355e-05)\n",
      "('statera', 8.210194818142367e-05, 1e-06)\n",
      "('Atreus', 7.9756178233383e-05, 6.432708504844532e-05)\n",
      "('Hermes', 7.9756178233383e-05, 1.1554514174740887e-05)\n",
      "('messe', 7.9756178233383e-05, 1e-06)\n",
      "('phantasia', 7.9756178233383e-05, 2.2109028349481774e-05)\n",
      "('metis', 7.741040828534232e-05, 1.0000000000000006e-06)\n",
      "('Nestor', 7.506463833730164e-05, 3.266354252422266e-05)\n",
      "('Gorgias', 7.506463833730164e-05, 7.488159922318621e-05)\n",
      "('Helene', 7.506463833730164e-05, 1e-06)\n",
      "('organo', 7.271886838926096e-05, 1e-06)\n",
      "('Troes', 7.037309844122029e-05, 1e-06)\n",
      "('Delos', 6.802732849317962e-05, 1e-06)\n",
      "('sto', 6.568155854513893e-05, 1e-06)\n",
      "('Europe', 6.333578859709826e-05, 1e-06)\n",
      "('Theodoro', 6.333578859709826e-05, 1e-06)\n",
      "('aude', 6.333578859709826e-05, 1e-06)\n",
      "('hei', 6.099001864905758e-05, 1.0000000000000006e-06)\n",
      "('tende', 6.099001864905758e-05, 1e-06)\n",
      "('exite', 5.86442487010169e-05, 1e-06)\n",
      "('elato', 5.86442487010169e-05, 1e-06)\n",
      "('emo', 5.86442487010169e-05, 1e-06)\n",
      "('nu', 5.629847875297623e-05, 0.001204214615920461)\n",
      "('Menelao', 5.629847875297623e-05, 1e-06)\n",
      "('Lemnos', 5.629847875297623e-05, 1e-06)\n",
      "('diametro', 5.629847875297623e-05, 1e-06)\n",
      "('Troas', 5.629847875297623e-05, 1e-06)\n",
      "('alueis', 5.395270880493555e-05, 1.0000000000000006e-06)\n",
      "('osse', 5.395270880493555e-05, 0.0001382086842716315)\n",
      "('lotos', 5.395270880493555e-05, 1e-06)\n",
      "('pe', 5.395270880493555e-05, 1e-06)\n",
      "('Zenoni', 5.1606938856894884e-05, 1e-06)\n",
      "('Phalereus', 4.9261168908854204e-05, 1e-06)\n",
      "('Eoi', 4.9261168908854204e-05, 1e-06)\n",
      "('eges', 4.691539896081353e-05, 1e-06)\n",
      "('Hesiodo', 4.691539896081353e-05, 1e-06)\n",
      "('Lesbo', 4.691539896081353e-05, 1e-06)\n",
      "('Pausania', 4.691539896081353e-05, 4.321805669896355e-05)\n",
      "('rhetori', 4.691539896081353e-05, 1e-06)\n",
      "('torno', 4.691539896081353e-05, 1e-06)\n",
      "('Sparte', 4.456962901277285e-05, 1e-06)\n",
      "('Mida', 4.456962901277285e-05, 2.2109028349481774e-05)\n",
      "('heroas', 4.456962901277285e-05, 1e-06)\n",
      "('Apolloni', 4.456962901277285e-05, 1e-06)\n",
      "('pompas', 4.456962901277285e-05, 0.0001593177126211133)\n",
      "('Eros', 4.222385906473218e-05, 1.0000000000000006e-06)\n",
      "('sei', 3.98780891166915e-05, 1.0000000000000006e-06)\n",
      "('athletas', 3.98780891166915e-05, 1e-06)\n",
      "('metri', 3.98780891166915e-05, 1.0000000000000006e-06)\n",
      "('siderea', 3.98780891166915e-05, 1e-06)\n",
      "('Naxos', 3.98780891166915e-05, 2.2109028349481774e-05)\n",
      "('hamos', 3.7532319168650824e-05, 1e-06)\n",
      "('Thebe', 3.7532319168650824e-05, 1e-06)\n",
      "('lita', 3.7532319168650824e-05, 1.0000000000000006e-06)\n",
      "('heroes', 3.7532319168650824e-05, 1e-06)\n",
      "('logo', 3.7532319168650824e-05, 1e-06)\n",
      "('Gorgia', 3.7532319168650824e-05, 3.266354252422266e-05)\n",
      "('Protagoras', 3.5186549220610144e-05, 1e-06)\n",
      "('ito', 3.5186549220610144e-05, 1e-06)\n",
      "('dele', 3.5186549220610144e-05, 1e-06)\n",
      "('antas', 3.5186549220610144e-05, 1.0000000000000006e-06)\n",
      "('eita', 3.5186549220610144e-05, 0.0027557281996073717)\n",
      "('Paros', 3.5186549220610144e-05, 1.1554514174740887e-05)\n",
      "('mone', 3.5186549220610144e-05, 1e-06)\n",
      "('Eous', 3.5186549220610144e-05, 1e-06)\n",
      "('elusa', 3.284077927256947e-05, 1.0000000000000006e-06)\n",
      "('polite', 3.284077927256947e-05, 1e-06)\n",
      "('Persephone', 3.284077927256947e-05, 1e-06)\n",
      "('lu', 3.284077927256947e-05, 1.1554514174740887e-05)\n",
      "('ambrosia', 3.284077927256947e-05, 1.0000000000000006e-06)\n",
      "('sidereos', 3.284077927256947e-05, 1e-06)\n",
      "('pare', 3.284077927256947e-05, 1.0000000000000006e-06)\n",
      "('Atrei', 3.0495009324528788e-05, 1.1554514174740887e-05)\n",
      "('doma', 3.0495009324528788e-05, 1.0000000000000006e-06)\n",
      "('manenti', 3.0495009324528788e-05, 1.1554514174740887e-05)\n",
      "('Parthenio', 3.0495009324528788e-05, 1e-06)\n",
      "('tropo', 3.0495009324528788e-05, 1e-06)\n",
      "('adamas', 3.0495009324528788e-05, 1.0000000000000006e-06)\n",
      "('aspis', 3.0495009324528788e-05, 0.00018042674097059507)\n",
      "('asto', 2.814923937648811e-05, 1e-06)\n",
      "('Lapithas', 2.814923937648811e-05, 1.0000000000000006e-06)\n",
      "('Minoa', 2.814923937648811e-05, 1e-06)\n",
      "('Leontinos', 2.814923937648811e-05, 2.2109028349481774e-05)\n",
      "('opta', 2.814923937648811e-05, 2.2109028349481774e-05)\n",
      "('diae', 2.814923937648811e-05, 1e-06)\n",
      "('Diones', 2.5803469428447435e-05, 1e-06)\n",
      "('mere', 2.5803469428447435e-05, 1.1554514174740887e-05)\n",
      "('stes', 2.5803469428447435e-05, 1e-06)\n",
      "('Cypria', 2.5803469428447435e-05, 1.0000000000000006e-06)\n",
      "('auge', 2.5803469428447435e-05, 1e-06)\n",
      "('problema', 2.5803469428447435e-05, 1e-06)\n",
      "('ano', 2.5803469428447435e-05, 1e-06)\n",
      "('Herme', 2.3457699480406762e-05, 1e-06)\n",
      "('beta', 2.3457699480406762e-05, 1e-06)\n",
      "('temo', 2.3457699480406762e-05, 1e-06)\n",
      "('Thebes', 2.3457699480406762e-05, 1e-06)\n",
      "('nomisma', 2.3457699480406762e-05, 0.0002437538260190404)\n",
      "('liga', 2.3457699480406762e-05, 2.2109028349481774e-05)\n",
      "('Naxo', 2.3457699480406762e-05, 1e-06)\n",
      "('pedali', 2.3457699480406762e-05, 1e-06)\n",
      "('problemata', 2.1111929532366085e-05, 1e-06)\n",
      "('nae', 2.1111929532366085e-05, 1e-06)\n",
      "('hero', 2.1111929532366085e-05, 1e-06)\n",
      "('Tantalo', 2.1111929532366085e-05, 1e-06)\n",
      "('elate', 2.1111929532366085e-05, 1e-06)\n",
      "('Arete', 2.1111929532366085e-05, 1e-06)\n",
      "('topo', 2.1111929532366085e-05, 1e-06)\n",
      "('iota', 2.1111929532366085e-05, 1e-06)\n",
      "('heroi', 1.876615958432541e-05, 1e-06)\n",
      "('Mimas', 1.876615958432541e-05, 1.0000000000000006e-06)\n",
      "('auges', 1.876615958432541e-05, 1e-06)\n",
      "('Euripo', 1.876615958432541e-05, 1e-06)\n",
      "('demo', 1.6420389636284732e-05, 1e-06)\n",
      "('sito', 1.6420389636284732e-05, 1e-06)\n",
      "('Amphitrite', 1.6420389636284732e-05, 1e-06)\n",
      "('Euripide', 1.6420389636284732e-05, 1e-06)\n",
      "('Same', 1.6420389636284732e-05, 1e-06)\n",
      "('periodo', 1.6420389636284732e-05, 1e-06)\n",
      "('mede', 1.6420389636284732e-05, 1e-06)\n",
      "('Epigenes', 1.6420389636284732e-05, 1.0000000000000006e-06)\n",
      "('geme', 1.6420389636284732e-05, 1e-06)\n",
      "('dogmati', 1.6420389636284732e-05, 1.1554514174740887e-05)\n",
      "('sape', 1.6420389636284732e-05, 1e-06)\n",
      "('emes', 1.6420389636284732e-05, 1e-06)\n",
      "('assa', 1.6420389636284732e-05, 1.0000000000000006e-06)\n",
      "('Hermogene', 1.6420389636284732e-05, 1e-06)\n",
      "('Miletos', 1.6420389636284732e-05, 1e-06)\n",
      "('agones', 1.6420389636284732e-05, 1e-06)\n",
      "('heroa', 1.6420389636284732e-05, 1e-06)\n",
      "('Melite', 1.6420389636284732e-05, 1e-06)\n",
      "('neo', 1.4074619688244056e-05, 1e-06)\n",
      "('aloes', 1.4074619688244056e-05, 1e-06)\n",
      "('Xanthippe', 1.4074619688244056e-05, 1e-06)\n",
      "('saturo', 1.4074619688244056e-05, 1e-06)\n",
      "('Delio', 1.4074619688244056e-05, 1e-06)\n",
      "('ese', 1.4074619688244056e-05, 1e-06)\n",
      "('Triptolemo', 1.4074619688244056e-05, 1e-06)\n",
      "('Aristophane', 1.4074619688244056e-05, 1e-06)\n",
      "('Leto', 1.4074619688244056e-05, 1e-06)\n",
      "('Protagora', 1.4074619688244056e-05, 1e-06)\n",
      "('Antiphates', 1.172884974020338e-05, 1e-06)\n",
      "('philo', 1.172884974020338e-05, 1e-06)\n",
      "('aloe', 1.172884974020338e-05, 1e-06)\n",
      "('polles', 1.172884974020338e-05, 1e-06)\n",
      "('Lede', 1.172884974020338e-05, 1e-06)\n",
      "('exerat', 1.172884974020338e-05, 1e-06)\n",
      "('Selene', 1.172884974020338e-05, 1e-06)\n",
      "('Laertes', 1.172884974020338e-05, 1e-06)\n",
      "('morio', 1.172884974020338e-05, 1e-06)\n",
      "('protero', 1.172884974020338e-05, 1e-06)\n",
      "('heroos', 1.172884974020338e-05, 1e-06)\n",
      "('ale', 9.383079792162704e-06, 1e-06)\n",
      "('aleis', 9.383079792162704e-06, 1.0000000000000006e-06)\n",
      "('lexeos', 9.383079792162704e-06, 1e-06)\n",
      "('mete', 9.383079792162704e-06, 1e-06)\n",
      "('exe', 9.383079792162704e-06, 1e-06)\n",
      "('lupe', 9.383079792162704e-06, 1e-06)\n",
      "('Hippotades', 9.383079792162704e-06, 1e-06)\n",
      "('hiet', 9.383079792162704e-06, 1.0000000000000006e-06)\n",
      "('meti', 9.383079792162704e-06, 1.0000000000000006e-06)\n",
      "('ares', 9.383079792162704e-06, 1e-06)\n",
      "('oles', 9.383079792162704e-06, 1e-06)\n",
      "('glossas', 9.383079792162704e-06, 1e-06)\n",
      "('ambrosio', 9.383079792162704e-06, 1e-06)\n",
      "('Melantho', 9.383079792162704e-06, 1e-06)\n",
      "('pedo', 7.037309844122028e-06, 1e-06)\n",
      "('trite', 7.037309844122028e-06, 1e-06)\n",
      "('metreta', 7.037309844122028e-06, 1e-06)\n",
      "('lae', 7.037309844122028e-06, 1.0000000000000006e-06)\n",
      "('lebes', 7.037309844122028e-06, 1e-06)\n",
      "('mese', 7.037309844122028e-06, 1e-06)\n",
      "('anathemata', 7.037309844122028e-06, 1e-06)\n",
      "('Ares', 7.037309844122028e-06, 1.0000000000000006e-06)\n",
      "('poesi', 7.037309844122028e-06, 1e-06)\n",
      "('Amphitrites', 4.691539896081351e-06, 1e-06)\n",
      "('bie', 4.691539896081351e-06, 1e-06)\n",
      "('arere', 4.691539896081351e-06, 1e-06)\n",
      "('etis', 4.691539896081351e-06, 1.0000000000000006e-06)\n",
      "('Polites', 4.691539896081351e-06, 1e-06)\n",
      "('Aretes', 4.691539896081351e-06, 1e-06)\n",
      "('erosa', 4.691539896081351e-06, 1e-06)\n",
      "('Theodore', 4.691539896081351e-06, 1e-06)\n",
      "('Pheres', 4.691539896081351e-06, 1e-06)\n",
      "('Eroti', 4.691539896081351e-06, 1e-06)\n",
      "('Laertiade', 4.691539896081351e-06, 1e-06)\n",
      "('psephismata', 4.691539896081351e-06, 1e-06)\n",
      "('bio', 4.691539896081351e-06, 1e-06)\n",
      "('Erote', 4.691539896081351e-06, 1e-06)\n",
      "('diastemata', 4.691539896081351e-06, 1e-06)\n",
      "('dine', 4.691539896081351e-06, 1e-06)\n",
      "('Melete', 4.691539896081351e-06, 1e-06)\n",
      "('Admete', 4.691539896081351e-06, 1e-06)\n",
      "('osa', 4.691539896081351e-06, 1e-06)\n",
      "('aule', 4.691539896081351e-06, 1e-06)\n",
      "('agonia', 4.691539896081351e-06, 1e-06)\n",
      "('Phemio', 4.691539896081351e-06, 1e-06)\n",
      "('Euenos', 2.345769948040676e-06, 1e-06)\n",
      "('Dardanide', 2.345769948040676e-06, 1e-06)\n",
      "('peribolo', 2.345769948040676e-06, 1e-06)\n",
      "('Diomedeos', 2.345769948040676e-06, 1e-06)\n",
      "('nassa', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('Parmenide', 2.345769948040676e-06, 1e-06)\n",
      "('Lampetie', 2.345769948040676e-06, 1e-06)\n",
      "('bolos', 2.345769948040676e-06, 1e-06)\n",
      "('Stesichorus', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('dromo', 2.345769948040676e-06, 1e-06)\n",
      "('parergo', 2.345769948040676e-06, 1e-06)\n",
      "('horai', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('pedalia', 2.345769948040676e-06, 1e-06)\n",
      "('Europes', 2.345769948040676e-06, 1e-06)\n",
      "('geometre', 2.345769948040676e-06, 1e-06)\n",
      "('Elpenor', 2.345769948040676e-06, 1e-06)\n",
      "('oleto', 2.345769948040676e-06, 1e-06)\n",
      "('hippo', 2.345769948040676e-06, 1e-06)\n",
      "('phthisis', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('axo', 2.345769948040676e-06, 1e-06)\n",
      "('Are', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('pale', 2.345769948040676e-06, 1e-06)\n",
      "('stephano', 2.345769948040676e-06, 1e-06)\n",
      "('Delon', 2.345769948040676e-06, 1e-06)\n",
      "('topazo', 2.345769948040676e-06, 1e-06)\n",
      "('Aristodemo', 2.345769948040676e-06, 1e-06)\n",
      "('metabole', 2.345769948040676e-06, 1e-06)\n",
      "('melo', 2.345769948040676e-06, 1e-06)\n",
      "('asbestos', 2.345769948040676e-06, 1.0000000000000006e-06)\n",
      "('Neritos', 2.345769948040676e-06, 1e-06)\n",
      "('pino', 2.345769948040676e-06, 1e-06)\n",
      "('phono', 1e-06, 1e-06)\n",
      "('nome', 1e-06, 1e-06)\n",
      "('xanthe', 1e-06, 1e-06)\n",
      "('agre', 1e-06, 1e-06)\n",
      "('pales', 1e-06, 1e-06)\n",
      "('pleto', 1e-06, 1e-06)\n",
      "('agrio', 1e-06, 1e-06)\n",
      "('ose', 1e-06, 1e-06)\n",
      "('geode', 1e-06, 1e-06)\n",
      "('daphne', 1e-06, 1e-06)\n",
      "('osi', 1e-06, 1e-06)\n",
      "('ites', 1e-06, 1e-06)\n",
      "('erato', 1e-06, 1e-06)\n",
      "('noma', 1e-06, 1e-06)\n",
      "('elateri', 1e-06, 1e-06)\n",
      "('Agamemnonides', 1e-06, 1e-06)\n",
      "('agoni', 1e-06, 1e-06)\n",
      "('prores', 1e-06, 1e-06)\n",
      "('melie', 1e-06, 1e-06)\n",
      "('apate', 1e-06, 1e-06)\n",
      "('alete', 1e-06, 1e-06)\n",
      "('lethe', 1e-06, 1e-06)\n",
      "('anterotas', 1e-06, 1e-06)\n",
      "('ereto', 1e-06, 1e-06)\n",
      "('ambrosie', 1e-06, 1e-06)\n",
      "('orsas', 1e-06, 1e-06)\n",
      "('hemeris', 1e-06, 1e-06)\n",
      "('thalle', 1e-06, 1e-06)\n",
      "('Megapenthes', 1e-06, 1e-06)\n",
      "('Hermionen', 1e-06, 1e-06)\n",
      "('aletes', 1e-06, 1e-06)\n",
      "('enen', 1e-06, 1e-06)\n",
      "('epistate', 1e-06, 1e-06)\n",
      "('Letoi', 1e-06, 1e-06)\n",
      "('iambe', 1e-06, 1e-06)\n",
      "('edos', 1e-06, 1e-06)\n",
      "('pedalio', 1e-06, 1e-06)\n",
      "('emata', 1e-06, 1e-06)\n",
      "('laro', 1e-06, 1e-06)\n",
      "('Neleus', 1e-06, 1e-06)\n",
      "('aniso', 1e-06, 1e-06)\n",
      "('orse', 1e-06, 1e-06)\n",
      "('depote', 1e-06, 1e-06)\n",
      "('aeto', 1e-06, 1e-06)\n",
      "('potho', 1e-06, 1e-06)\n",
      "('hebe', 1e-06, 1e-06)\n",
      "('telamon', 1e-06, 1e-06)\n",
      "('polle', 1e-06, 1e-06)\n",
      "('psoras', 1e-06, 1e-06)\n",
      "('admete', 1e-06, 1e-06)\n",
      "('doto', 1e-06, 1e-06)\n",
      "('dido', 1e-06, 1e-06)\n",
      "('zeta', 1e-06, 1e-06)\n",
      "('peren', 1e-06, 1e-06)\n",
      "('daphnes', 1e-06, 1e-06)\n",
      "('meni', 1e-06, 1e-06)\n",
      "('pege', 1e-06, 1e-06)\n",
      "('halo', 1e-06, 1e-06)\n",
      "('hesperos', 0.0, 2.2109028349481774e-05)\n"
     ]
    }
   ],
   "source": [
    "shared_latin_greek = [(word, \n",
    "                       latin_word_probs.get(word, 1e-06),\n",
    "                       greek_transliterated_word_probs.get(word, 1e-06)) \n",
    "                      for word in shared_words]\n",
    "\n",
    "shared_latin_greek.sort(key=lambda a: a[1], reverse=True)\n",
    "for item in shared_latin_greek:\n",
    "    print(item)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`kai` is the most common word in the Greek corpus, so we could also divide the shared words by the threshold of this probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0011142407253193212"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "latin_word_probs.get('kai')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "97\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['suos', 'at', 'pares', 'has', 'eris', 'esse', 'de', 'deos', 'ea']"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "likely_latin = [word for word, latin_prob, greek_prob \n",
    "                in shared_latin_greek \n",
    "                if latin_prob >= latin_word_probs['kai']]\n",
    "print(len(likely_latin))\n",
    "random.sample(likely_latin, 9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "487\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['arere',\n",
       " 'Euripo',\n",
       " 'pio',\n",
       " 'Europes',\n",
       " 'dote',\n",
       " 'agro',\n",
       " 'Hesiodo',\n",
       " 'Stesichorus',\n",
       " 'agre']"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "greater_prob_latin = [word for word, latin_prob, greek_prob\n",
    "                     in shared_latin_greek\n",
    "                     if latin_prob >= greek_prob]\n",
    "print(len(greater_prob_latin))\n",
    "random.sample(greater_prob_latin, 9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "39,885 distinct transliterated Greek words without matches in the Latin corpus\n"
     ]
    }
   ],
   "source": [
    "# We'll remove the words that have a high probability of being Latin \n",
    "# from the collection of demacronized transliterated Greek words\n",
    "only_greek_transliterated = distinct_demacronized_greek - set(likely_latin)\n",
    "likely_greek = shared_words - set(likely_latin)\n",
    "\n",
    "# Likewise, let's remove the transliterated words that are likely Greek \n",
    "# from the collection of good Latin words\n",
    "distinct_good_latin = set(distinct_good_latin) - likely_greek\n",
    "\n",
    "print(f'{len(only_greek_transliterated):,} distinct transliterated Greek words without matches in the Latin corpus')\n",
    "\n",
    "# NOTE: we are toggling this on to see the difference\n",
    "# only_greek_transliterated=  distinct_demacronized_greek\n",
    "# distinct_latin_wo_greek_matches = distinct_good_latin - shared_words\n",
    "# The following had low precision high recall:\n",
    "# only_greek_transliterated = distinct_demacronized_greek - shared_words"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Normally these gaps might concern us, but since we are more interested in groups of loanwords in phrases, we can rely on smoothing over clusters of loanwords to screen out misses."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create a simple data matrix of the single words, transliterated Greek examples followed by the Latin words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "219377"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = [list(only_greek_transliterated) + list(distinct_good_latin)]\n",
    "len(X[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Before we featurize our data matrix, let's check on the max word lengths"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "28 Max word length in distinct good Latin sample\n",
      "24 Max word length in transliterated Greek sample\n"
     ]
    }
   ],
   "source": [
    "max_latin_len=sorted([len(tmp) for tmp in distinct_good_latin])[-1]\n",
    "max_greek_len =sorted([len(tmp) for tmp in only_greek_transliterated])[-1] \n",
    "max_len =max(max_latin_len, max_greek_len)\n",
    "print(f'{max_latin_len} Max word length in distinct good Latin sample')\n",
    "print(f'{max_greek_len} Max word length in transliterated Greek sample')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "x_train, x_test, y_train, y_test = train_test_split(all_X, all_y, test_size=0.30)\n",
    "x_test, x_validation, y_test, v =train_test_split(x_test, y_test, test_size=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()\n",
    "# try:  all_X.astype(np.float64)\n",
    "# Fit scaler only on training slice of data; \n",
    "# As with all the transformations, it is important to fit the scalers to the training data only,\n",
    "# not to the full dataset (including the test set).\n",
    "scaler.fit(X_train.astype(np.float32))\n",
    "# TODO pickle this scaler for applying the same transforms to input\n",
    "with open('scaler.pkl', 'wb') as fwriter:\n",
    "    pickle.dump(scaler, fwriter)    \n",
    "# Fit all 3 data sets\n",
    "X_train = scaler.transform(X_train.astype(np.float32))\n",
    "X_test = scaler.transform(X_test.astype(np.float32))\n",
    "X_validation = scaler.transform(X_validation.astype(np.float32))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Featurization\n",
    "#### We'll use a generic character to integer transform so that we can reuse our encoding process for unseen character data combinations (as opposed to building a dictionary mapping for a discrete sample space)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Let's create the X,y feature matrix and labels and save them for re-use in other notebooks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y shape: (219377,)\n",
      "X shape: (219377, 28)\n"
     ]
    }
   ],
   "source": [
    "all_y = np.array([1] * len(only_greek_transliterated) + [0] * len(distinct_good_latin), dtype=int)\n",
    "print(f'y shape: {all_y.shape}')\n",
    "with open('all_y', 'wb') as fwriter:\n",
    "    pickle.dump(all_y, fwriter)\n",
    "# We use a label encoder to automatically capture the range of values for provenance\n",
    "# Although it's true we're only doing binary classification, \n",
    "# it's a good practice to use this so that we can automate recording our model's provenance\n",
    "label_encoder = LabelEncoder()\n",
    "label_encoder.fit(all_y)\n",
    "all_words = list(only_greek_transliterated) + list(distinct_good_latin)\n",
    "all_X = np.array([\n",
    "#     word_to_features(word, max_len) \n",
    "    encode_word(word, the_char_idx, max_len)\n",
    "    for word in all_words])\n",
    "print(f'X shape: {all_X.shape}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(all_X, all_y, test_size=0.30)\n",
    "X_test, X_validation, y_test, y_validation =train_test_split(X_test, y_test, test_size=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standard scalar\n",
    "scaler = StandardScaler()\n",
    "# try:  all_X.astype(np.float64)\n",
    "# Fit scaler only on training slice of data; \n",
    "# As with all the transformations, it is important to fit the scalers to the training data only,\n",
    "# not to the full dataset (including the test set).\n",
    "scaler.fit(all_X.astype(np.float32))\n",
    "# TODO pickle this scaler for applying the same transforms to input\n",
    "with open('scaler.pkl', 'wb') as fwriter:\n",
    "    pickle.dump(scaler, fwriter)    \n",
    "# Fit all 3 data sets\n",
    "X_train = scaler.transform(X_train.astype(np.float32))\n",
    "X_test = scaler.transform(X_test.astype(np.float32))\n",
    "X_validation = scaler.transform(X_validation.astype(np.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('all_X.pkl', 'wb') as fwriter:\n",
    "    pickle.dump(all_X, fwriter)\n",
    "all_X = sparse.csr_matrix(all_X)\n",
    "with open('all_X_scaled.pkl', 'wb') as fwriter:\n",
    "    pickle.dump(all_X, fwriter)\n",
    "num_samples = all_y.shape[0] # to be used later by model provenance\n",
    "num_features = all_X.shape[1] # to be used later by model provenance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train a DummyClassifier to show the baseline which we must improve above"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dummy classifier: 0.7043303856322363\n"
     ]
    }
   ],
   "source": [
    "dummy = DummyClassifier(strategy='stratified', random_state=0)\n",
    "features_train, features_test, target_train, target_test = train_test_split(all_X, all_y,\n",
    "                                                                            random_state=0)\n",
    "dummy.fit(features_train, target_train)\n",
    "dummy_score = dummy.score(features_test, target_test)\n",
    "print(f'Dummy classifier: {dummy_score}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train and classify the data using several classifiers, printing out the cross validation score results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  0%|          | 0/4 [00:00<?, ?it/s]\u001b[A\n",
      " 25%|██▌       | 1/4 [08:16<24:49, 496.40s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classifier type: <class 'sklearn.neighbors.classification.KNeighborsClassifier'> mean score: 0.8701504715049067 scores: 0.8694958519463943,0.8708177591393929,0.8702222222222222,0.8711111111111111,0.8691054131054131\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      " 50%|█████     | 2/4 [08:58<12:00, 360.02s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classifier type: <class 'sklearn.ensemble.bagging.BaggingClassifier'> mean score: 0.9579035273455183 scores: 0.9577673443340323,0.9559212325644999,0.9596125356125356,0.9567635327635328,0.9594529914529915\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      " 75%|███████▌  | 3/4 [09:23<04:19, 259.51s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classifier type: <class 'sklearn.ensemble.forest.ExtraTreesClassifier'> mean score: 0.9417486903651551 scores: 0.940035554745191,0.941015589388276,0.9431111111111111,0.9397606837606838,0.9448205128205128\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "100%|██████████| 4/4 [09:41<00:00, 145.32s/it]\u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classifier type: <class 'sklearn.ensemble.forest.RandomForestClassifier'> mean score: 0.95025003767416 scores: 0.9480353724131644,0.9499498586926793,0.9502678062678063,0.9495612535612535,0.9534358974358974\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "classifiers = [\n",
    "    KNeighborsClassifier,\n",
    "    BaggingClassifier,\n",
    "    ExtraTreesClassifier,\n",
    "    RandomForestClassifier\n",
    "]\n",
    "for cls in tqdm(classifiers):\n",
    "    scores = cross_val_score(cls(), all_X, all_y,\n",
    "                             scoring='accuracy',\n",
    "                             n_jobs=multiprocessing.cpu_count(),\n",
    "                             cv=5)\n",
    "    print(f\"Classifier type: {str(cls)} mean score: {scores.mean()} scores: {','.join([str(tmp) for tmp in scores])}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "oob score: 0.9626 x_test score 0.9625 validation score 0.9612\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300,\n",
    "    bootstrap=True, class_weight=None, criterion='gini',\n",
    " max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
    " min_impurity_decrease=1e-07, min_samples_leaf=1,\n",
    " min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
    " n_jobs=-1, oob_score=True, random_state=None,\n",
    " verbose=0, warm_start=False)\n",
    "rf.fit(X_train, y_train)\n",
    "y_pred = rf.predict(X_test)\n",
    "print(f\"oob score: {rf.oob_score_:.4f} x_test score {rf.score(X_test, y_test):.4f} validation score {rf.score(X_validation, y_validation):.4f}\")\n",
    "# 100\n",
    "# 300 estimators\n",
    "# oob: 0.9547221661467932, x_test 0.9544777706870878 0.9559972042422585 roc_auc 0.8880479620121011"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9083355407702502"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, y_pred)\n",
    "roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "roc_auc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(false_positive_rate, true_positive_rate, label='Classifier')\n",
    "plt.plot([0,1], ls='--', label='No skill')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver operating characteristic')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.96346522, 0.96385268, 0.96540171, 0.965151  , 0.96658689])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "score= cross_val_score( RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
    " max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
    " min_impurity_decrease=1e-07, min_samples_leaf=1,\n",
    " min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
    " n_estimators=300, n_jobs=-1, oob_score=False, random_state=None,\n",
    " verbose=0, warm_start=False) , all_X, all_y,\n",
    "                             scoring='accuracy',\n",
    "                             n_jobs=multiprocessing.cpu_count(),\n",
    "                             cv=5)\n",
    "score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# If you want you can train and graph the effects of different parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbUAAAEfCAYAAADGLVhVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdeXxMV//A8c/MZI9ELAlFQhHEVtRW1VKUoii1e9RSSyu1r63dQ9GUWh5FpanGUqJiq/ZRD+nGL12UtigSS1URaTQhyUxmu78/UuHKYqJJJjP5vl+veSVzzl2+J5OZ79x7zzlXk5ycrCCEEEI4Aa29AxBCCCEKiiQ1IYQQTkOSmhBCCKchSU0IIYTTkKQmhBDCaUhSE0II4TQkqQkhhHAaktSEEEI4jRKZ1OLi4uwdQqGRtjkmaZvjcub2OWLbSmRSE0II4ZwkqQkhhHAaktSEEEI4DbsmtSNHjtC/f39CQkLw8/Njy5YtD1zn1KlTdOnShYoVKxISEsLSpUtRFJmTWQghBLjYc+dpaWnUrVuXAQMG8Morrzxw+Vu3btGzZ09atWrF4cOHiYuLIzQ0FC8vL8aOHZvv/ZvNZtLS0h4m9GLLw8ODlJQUe4dRKGxtm7e3Ny4udv3XFkLYiV3f+R07dqRjx44AjBkz5oHL79ixA71ez9q1a/H09KRu3bqcO3eOd999l9deew2NRmPzvs1mM7dv38bPzy9f6xV37u7ueHh42DuMQmFL2xRFITk5GR8fH0lsQjwERQGjETIyIDlZx/XrGkwmMJszf9593H1uNmuwWMBqvfuwWDK3ZbVqsp5brVCpkpU2bSyFFr9Dveu/++47nnjiCTw9PbPK2rdvz6JFi/jtt9+oVq2azdtKS0tzuoQmQKPR4Ofnx61btyhdurS9wxGiUFkskJKi+ftx7+85P27d0pCRARkZmr8TlyYrgRmNmXUm072fiY0LPObnnjPRpk16gW/3DodKajdu3KBSpUqqMn9//6y6/CQ1QBKak5LXVTgqRYGrVzXExWlJStJy86aGmzc1/PXX3Z93fr95MzNRKYpj/b8XdhcIh0pqDyunAYS3bt3C3d3dDtEUPoPBYO8QCo2tbbt16xY3btwo5GgKliMOdLWVM7cNHq596elaLlzwJC7Ok/h4T+LjvYiP9+TWLef+WL59O/0f/z8EBwfnWudQf72AgAASExNVZXeeBwQE5Lre/X+AuLg4fH19nfLak8FgcMp2Qf7a5uvrS2BgYCFHVHDi4uLyfKM6MmduGzy4fenpcOmSlrg4LadP6zh1SsepU1ouXtQVYZT54+am4OYGOp0Fd3ctrq7g4gKurpnld36/t1ynA61W/cgsU1Rl9eu7Fer/g0MltebNmzNv3jzVh1tMTAyPPPIIVatWtXN0jm/48OGYzWYiIyPtHYoQDiU9HS5e1HL+vPbvnzouXMj8/Y8/CnfklK+vgp+fQunS2R++vtmfe3pmJi13d3B3v/v73Z9w5wy+I34hsWtSS01N5cKFCwBYrVauXLnCzz//TJkyZQgMDGT+/PkcO3aMvXv3AtC7d2+WLl3KmDFjmDJlCvHx8axYsYJp06aViOsofn5+edYPGDCAd95556G3/8477xTomL+mTZty6dIlfvzxR4KCglR1HTp0oGXLlixcuFBVHhERwcKFC7P+LwCSk5NZuXIl+/bt4/fff6d06dLUqVOH4cOH0717d7RamUNAFK3LlzVERbnxxRcunDvXkBs33Aps297eCiEhFipVUihb1krZsgplyiiqn3cefn4K0slXza5/juPHj9OtW7es54sXL2bx4sUMGDCAtWvXcv36dS5evJhVX7p0aXbt2sWUKVN45pln8PPzIzQ0lNdee80e4Re5s2fPZv1+4MABxo0bpyrL7dScyWTC1dX1gdsvyN6CR44cISMjg169erF582beeOONh9pOUlISnTp1wmAwMG3aNJo3b46LiwtHjx5l4cKFPPHEE1SoUKHA4hYiN7duwd69rmzb5sY33/zzj06NRqF6dSv16lmpV89C3boW6te3UrWqFfme9vDsmtSeeuopkpOTc61fu3ZttrJ69erx2WefFWZYxda9H953EtD9H+i//PILTz31FBs3buS9997j2LFjhIWF8dxzzzF9+nRiY2NJTk6mWrVqTJw4kb59+2ate//pxw4dOtCsWTNcXFzYtGkTbm5uDBo0iDlz5jzwyHjTpk307duX1q1bM3bsWGbMmPFQR1Rz5swhISGBH374gdKlS2cl7po1a9K7d28ZiyYKldkMX37pwrZtrnzyiSt6ff7PCGk0CoGBCo8+aqVOHQv16mUmr9q1LXh7F0LQJZx8IjyAn1/BHb0kJxfdTB/z5s1j4cKF1K9fH3d3d/R6PU2bNmXixIn4+vpy8OBBXn31VQIDA3niiSdy3c6WLVsYO3Yshw4d4tixY7zyyis0adJEdYR9v5SUFPbu3UtMTAzBwcFYLBYOHz5Mhw4d8tUGs9nM7t27GTRoEBUqVMjW89HLyytf2xPCVidPatm2zY2PP3bl+vUHfxnTahWCgqxUr579UbWqFSftaF0sSVJzUqGhoTz//PPZyu4YOXIkMTExREdH55nUGjZsyNSpUwGoUaMGH3zwAV999VWeSW3nzp3UqlWL2rVrA9CnTx8iIyPzndSuXbtGWlpa1naEKEzJybBtmxubN7tx8uSDeyY2bWqmf38TVavG06ZNEG4Fd1lN/AOS1JxU48bqmQDMZjNvv/02e/fu5erVq5hMJjIyMh6YaOrVq6d6XrFixWzDKu4XGRlJv379sp7369ePtm3b8ueff1K+fHmb2yATVYui8OOPOiIi3Ni588GnF6tUsdK/v5F+/UwEB1sBiIvLkIRWjEhSc1L3n5p7++23CQ8PZ/HixdSpUwdvb29mzpyJ0WjMczv3dzDRaDRYLLnP2/bLL79w4sQJfv75Z2bNmpVVbrFY+Oijj7Imnvbx8eHWrVvZ1k9JScHX1xeASpUq4e3treoMI0RBSEuDnTtdiYhw48SJvD8GfXwUevQw0b+/kVatLNKJo5iTpPYARXkdrDDFxsby/PPP06dPHyBzCMX58+cLfIByZGQkrVq14u2331aVf/zxx2zevDkrqdWsWZNvv/022/o//fRT1rgYFxcXevTowUcffcTkyZOz9c5MT0/HxcUFN/maLGx05oyWiAg3tm1z49at3I/KtFqFdu0yTy926WJCLt86DvnOUULUrFmTQ4cO8d1333H27FkmTJjA9evXC3QfBoOBHTt20LdvX+rWrat6DBkyhLNnz2YlslGjRvHrr7/yxhtvcOrUKc6dO5c1Fu3eIRoLFizA39+f9u3bExUVxZkzZzh//jxbtmyhTZs2/PXXXwXaBuF8MjIyj8q6dPGmZUsf3nvPPdeEVqWKlZkzDZw6dZuPP06nd29JaI5GjtRKiNdff50rV67Qs2dPvLy8eOmll+jevTvXrl0rsH3s27eP27dvZ+ugAlC1alWaNGlCZGQkLVq0IDg4mP3797No0SK6d++OyWSidu3aWcnqjvLly3P48GHeeecdVq5cyZUrV/D19aVOnTrMmjUrz+nRRMl28aKWjRvd2LLFlT//zP37u0aj8OyzZoYNM9Kxoxld8Z29SthAk5ycXOKuxsfFxREQEOCUtyaRuR8zpaSkONTr64jTEdmqKNtmMsGnn7rwwQdufPFF3hMO+PtbGTzYyEsvGalW7eE/BuW1K17kSE0I4fAuX9YQGenGpk1uJCTkfVXlySfNDB9upFs3k/RadEKS1IQQDslshgMHXNi40Y3//c8lz/uK+foq9O9vZPhwI3XqWIswSlHUJKkJIRyKyQTr1rmxdq07V6/mfVTWtKmZoUON9OolHT5KCklqQgiHcfq0ljFjPPMcW1aqlELfvkaGDjXSsKEclZU0ktSEEMWe2QyrV7uzeLE7RmPOpxkfe8zC8OEZvPiiiVKlijhAUWxIUhNCFGvnzmUenf3wQ/aPKy8vhd69TQwbZqRx49xnuhElhyQ1IUSxZLHAu++6sXChBxkZ2Y/OevQwsWyZnvLlS9yoJJEHSWpCiGLn/HktoaGexMZm/4gqU8bKsmUGevUy2SEyUdzJNFlCiGLDaoX1691o3bpUjgmtSxcTsbGpktBEruRITQhRLFy6pOG117z45pvsH0ulSyuEhenp08fEA266Lko4OVJzIH5+fnk+Xn311X+8j3PnzuHn58fp06dtXmf8+PGULVuWbdu2ZaubN28ebdu2tWk/VquViIgI2rdvT+XKlQkKCqJNmzb85z//ITU19aHaIxzD9u2utG7tk2NC69jRRGzsbfr2lYQmHkyO1BzIvfcVO3DgAOPGjVOV2WPOx7S0NHbt2sWECROIjIykf//+D7UdRVEYNmwYn3/+OVOnTiUsLIxy5cpx+vRp1q1bxyOPPELXrl0LOHphb7dvw9Spnmzbln2+Kl9fhTff1DNokCQzYTtJag6kQoUKWb/fmaz33jLInPT38uXLzJ49m5iYGLRaLS1btmTJkiVUq1YNgEuXLjF9+nRiY2MxmUwEBgYyc+ZMOnbsSPPmzQFo1aoVAO3bt2fnzp25xrRr1y5CQkKYMGECtWvX5vz589SoUSPfbdu2bRt79uwhKiqKjh07ZpVXrVqVzp07k5ycnO9tiuLtxAktL7/sxfnz2afFb9fOxKpVeqpUkZ6NIn8kqd2ntJ9fke4vpYA/rFNTU3n++ed55pln+Oyzz3BxcWH58uX07NmT2NhY3N3dmTBhAm5ubnz66ad4e3tz7tw53Nzc8PDw4LPPPqNz58588sknBAcH4+7unuf+Nm3aRN++ffH19aVz585s2rSJefPm5TvuHTt2ULduXVVCu5efnx8GgyHf2xXFj9Wa2VV//nwPTCb1IZiHh8KiRQaGDzfK0Zl4KHJNzcns2LEDb29vVq5cSb169ahduzb/+c9/SExM5NChQwD8/vvvtGrVinr16lGtWjU6duyYdd2rXLlyAJQtW5YKFSrgl0eSP3fuHMePH6dXr14A9O/fn48++giz2ZzvuM+fP0/t2rXzvZ5wLImJGvr182LWLM9sCS0kxMLhw6m8/LIkNPHwJKk5mZ9//plz585RuXLlrEe1atVIS0vj4sWLALz66qssXLiQTp06sWjRIn7++eeH2ldkZCTPPvssZcqUAaBdu3YoisKBAwfyvS1FkdNMzu6LL3S0bl2Kgwez3+ds2LAMDh1KpW5dmatR/DNy+tHJWK1WmjZtytq1a7PVlS1bFoARI0bQqVMnDh48yBdffMGKFSt44403mDhxos37MZlMbNu2jaSkpKyjuzv7j4yMzOrU4ePjw61bt7Ktn5KSAoCvry8ANWvWVHV6Ec7DbNYwf747K1a4Z7s9TOnSCqtWpdOjR/6P7oXIiSS1+xT0Na6i1qBBA/73v//h7++Pj49PrssFBgYyfPhwhg8fzpIlS9i4cSMTJ07E7e+7Jlosec+j9+mnn2IwGPjqq6/Q6e5e6D9//jxDhw7l2rVrPPLII9SsWZMrV65kS34//fQT3t7eVK5cGYDevXszZswYPv/88xyvqyUnJzvtHb2d2aVLGkaOrM3Jk9lfu5YtzWzYkE5goByli4Ijpx+dTL9+/ShVqhSDBg3i6NGjXLp0iW+++Ybp06dz+fJlAKZOncrhw4e5dOkSJ06c4IsvvqBOnTpAZm9KNzc3Dh06RGJiYo5HWZDZQeS5556jQYMG1K1bN+vRrVs3KleuzNatWwHo3LkzgYGBvPzyy3z33XdcunSJXbt2sXTpUkJDQ9H8ffGkf//+dO3alSFDhrB8+XKOHz/O5cuXOXDgAD179sy6Higcx/HjOtq08eHkSfWU+RqNwtSpBj75JE0SmihwktScjI+PD//973+pWLEigwcPpkWLFoSGhqLX67NO9ZlMJiZNmkSLFi3o3bs3gYGBrF69GgBPT0/efPNNNmzYQO3atRk2bFi2fVy5coXDhw/TvXv3HGN44YUX2Lx5M4qi4Obmxp49ewgICGD48OG0atWKt956i3HjxjFt2rSsdbRaLZs2bWLBggXs27ePrl270rp1a958803at2/Pc889Vwh/LVFY4uO19O7tRUqK+nRjpUpW9u5NY+bMDFzkPJEoBJrk5OQS91UpLi6OgICArLFezsRgMDjtabr8tC0lJcWhXt+4uDiCg4PtHUaBuH5dQ8eOpbh8Wf2d+bnnTKxZo6dcOef6yHGm1+5+jtg2+a4khCgwKSnw4ove2RLauHEZzJ9vkK76zuLWLXTnz6ONi8t8/PYbmoyMzDpFUT/uK7M0aULGPWdpCpokNSFEgTAYYOBAb06dUs8Q0q3bn8yf7yoJzd4UBU1ycuaN6rRaFK0W8npYLLj/9hsufycvXXw82juPhIR/FEdhkqQmhPjHLBYYMcKLI0fUHymdOpl4441LaDSOdQrLIVmtaBIS0F6+jPb339Fevozm75/a339H+/vvaPT6fG2yQWHE6exJLTw8nFWrVpGQkECdOnVYvHhx1ryDOdmwYQMbNmzg8uXLVKlShcmTJzNgwIAijFgIcS9FgcmTPfjkE/Wg6ubNzXzwQTp//GGnwByVoqC5cQNNcjKa27fR3LoFt2+jSUlBc+tWVlnWz+RkNFeuoL1yBY3RaO/o7c6uSS06OpoZM2awbNkyWrZsSXh4OH369CE2NpbAwMBsy7///vvMmzePlStX0rRpU44dO8b48ePx8/Ojc+fOdmiBEGLJEnc2blTPEVq7toXt29Px8rJTUI7AYkF76RLas2fRnT2L9uxZtOfOoTt3Dk0xv9WS4uKC9dFHsdasibVmTSw1a8LfvasVjYasc805/K4EBBRqbHZNamvWrGHgwIEMGTIEgLCwMA4dOkRERARz587Ntvz27dt56aWX6N27NwDVqlXjxx9/ZOXKlQ+V1BRFyRonJZyHTLlVdN5/342lS9U9UitXtrJzZxplysjrAMDt22gvXEB34UJm4rqTxM6fv9u5oogo3t4o7u6Zpyqt1szDbKs1x4dGUTCWK4e2Tp2sxGUNDs58BAWBa/bpzooDuyU1o9HIiRMnGDt2rKq8Xbt2fPvttzmuk5GRka1Lt6enJ8eOHcNkMuGajz+yt7c3ycnJ+Pn5SWJzIoqikJycnOdsKqJg7NnjwpQp6vdjmTJWoqPTSt4tY5KT0V28iPbChbuPO89v3CiyMKx+fihBQVgDAzMfd34PCkIJCkLx88PmHjuKQlx8vHTpt1VSUhIWiwV/f39Vub+/Pzdy+Sdo3749mzZtolu3bjRu3JgTJ04QGRmJyWQiKSmJihUr5rheXFxctrI7k/veunULrVbGoDsLq9WK0Wjkzz//tHco+ZbT/2lx9cMPPowbF6yay9Hd3cLbb59Dq03j/qY4UtseSFHwOn2aMocP43P8OI9dvozr33OZFhSLtzemcuWweHtjKVXq7s97f7+nzOTvT0bFilhLlcp9o3/+mfnIp+L42uWVaO3eUSQ/pk6dSkJCAh07dkRRFAICAhgwYAArV67MMzHd/wdwxAGFtpK2OSZHattPP2mZNq2U6tYxOp1CZKSBTp0qZVvekdqWK6sV3bff4rp3L6779qG9cqVgNuvvj7V2bSy1a2OtVQtLnTpYa9VCqVhRdUSl/ftR1Cf8HPG1s1tSK1euHDqdjsTERFV5YmIiAblcSPT09GTNmjWsWLGCGzduULFiRTZu3IiPjw/ly5cvirCFKNFOnNDSt683t2+rT2GtXq2nUycnm2nfbEZ35EhmIvvkk4cem6W4uGCtVi2zY0VwcGYC+/uh/H3bJlFw7JbU3NzcaNSoETExMbzwwgtZ5TExMbnOKXiHq6tr1uzuO3fupFOnTnIKUYhCtnmzK5Mne5KRoU5o8+frGTjQZKeoClh6Oi7ffIPr3r24fPop2ps3bVpNcXPLTFrVq6selkcfRalSBZnosujY9S8dGhrK6NGjefzxx2nRogURERFcv349axLd0aNHA7B+/XoA4uPj+eGHH2jWrBnJycmsWbOGX3/9Ncd7hwkhCobRCDNmeBAR4Z6tbsyYDMaNc9CxUQYDupMn0Z04ge74cXTHj6M9cyazV+ADKJ6emDt0wNS9O/EVKlD1ySdBp3vgeqLw2TWp9erVi5s3bxIWFkZCQgIhISFERUURFBQEZM4Gfy+LxcKaNWuIj4/H1dWV1q1b8/nnn1O1alV7hC+E07t6VcOQIV58/332j4pXXslg4UIHmc/RaER7+jQufycv3fHjaH/9FY3Z9lOmio8Ppk6dMHXrhrlDB/D2ztx0XJwktGLE7sfEI0aMYMSIETnW7d+/X/W8du3afP3110URlhAl3pEjOoYO9SIxUX1q39NTYcUKPf36FfNTjsnJuH7yCa7R0bh8881DzbZh9fPD3KULpu7dMbdtC056BwxnYvekJoQoXhQF1q51Y/ZsDywW9WFY1apWNm9Oo0GDB5+is4u0NFz/+19cd+7E5X//e6hEZnn0Ucxt2mDu0QNz69bFdpCxyJkkNSFElrQ0GD/ek48/dstW16GDiQ0b9MVvppCMDFwOHsQ1OhrX//4XTXq6zatag4KwNG6MuXFjLI0bY3nsMfDzK8RgRWGTpCaEAODCBS3/+pcXp09nvz40daqBGTMyis+lI7MZly+/xHXnTlw/+SRz0t8HsFaqlJm47jwaNUIpV64IghVFSZKaEIIDB1wYOdKLW7fUpxt9fRXWr0+nc+diMgZNUXCNjsZj/ny0ly8/cHFL9eqYXnwRU69eWENCiiBAYW+S1IQo4fbscWHoUC/VlFcAISEWNm9Op0aN4nH9TBcbi8fMmbgcO5bnctYqVTD17InxxRexPvaY7XMdCqcgSU2IEuzECS2vvJI9ofXqZWTVKj15TSVYVLQXLuAxbx6ue/fmuozV3x/TCy9gevFFLM2bZ965WZRIktSEKKGuXdMwcKA3er16Dsf58w2EhhrtfoCj+esv3N96C7fwcDSm7MMHFBcXTH37YuzbF0vr1jJrhwAkqQlRIun1MGiQF1evqo9o3nlHz0sv2Xn8WUYGbhs24BEWhiaX2e9Nzz+PYd48rDVrFnFworiTpCZECaMoEBrqyY8/qt/+oaEZ9k1oioLLnj14zJuH7tKlHBcxN26MYeFCLE8+WbSxCYchSU2IEuatt9yJjlaPQ+vY0cSCBQb7BJScjNvHH+MWGYnu559zXMRapQqGuXMxvfiiXC8TeZKkJkQJsnu3C4sXq6d6CgmxEB6eXrRj0CwWXL76CtfNmzPHmWVk5LiY4uNDxqRJZLzyCnh6FmGAwlFJUhOihDhxQsurr3qpysqVs/LRR2n4+hZNDNqLF3HdsgW3bdvyvNGmotNhHDaMjOnTUfz9iyY44RQkqQlRAly7pmHAAHVPR1dXhU2b0qlWrXCnvdLq9bh+9BFumzfjcuRInssqGg3mzp0xzJ2LtXbtQo1LOCdJakI4ufR0GDjQi2vX1Neili/X06qVpfB2bDDgsXAhj33wAbq0tDwXtVatinHQIIwDBqAEBhZeTMLpSVITwond6el4/Lj6rf7aaxkMHlx4PR01N27gNWgQLt9/n3tsnp6YevTAOGhQZm9G6QAiCsBDJbXz58+TmJhISEgIpUuXLuiYhBAFZOlSd3btUvd07NTJxPz5hdfTUXvyJN79++d6zczcvDnGf/0L0wsvUGQX80SJka+vRjt27KB+/fo0a9aMLl26cOLECQCSkpJ4/PHH2bVrV6EEKYTIv127XFmyJHtPxw0bCq+no8tnn1HqueeyJTRrQAAZ48dz+7vvSPv8c0wvvSQJTRQKm5Panj17GDVqFLVq1WLBggUoyt2Ly+XKlaNWrVps27atUIIUQuRPZk9HdRf4Qu3pqCi4rV6N18CBaFJTVVUpLVpw+7vvMMyfj7VWrULYuRB32ZzUli1bRtu2bYmOjmbgwIHZ6ps2bcrJkycLNDghRP7duKFh0CBvDIYi6uloNOI5bhyes2ejUdTbzxgxgrgVK+TGm6LI2JzUzp07x/PPP59rvb+/P3/++WeBBCWEeDhGI7z0khd//JF9TsfC6OmouXkT7549cdu0SVWu6HTow8IwvP22TDQsipTN/21eXl6k5dEt9+LFi5STu8gKYVfTp3sQG6t+W48Zk8G//lXwPR21587h1a8fuosXVeWKry/pGzdibteuwPcpxIPYfKT29NNPs3XrVoxGY7a6a9eu8eGHH9JO/omFsJsPPnDjgw/cVWVt2xbOnI4uMTGU6tAhW0KzVKtG6sGDktCE3dic1GbPns3169dp27Yt4eHhaDQaDh48yLx582jVqhVarZbp06cXZqxCiFwcPapj6lR1T8eqVa1EROgL9uyfouAWHo5X795obt1SVZlbtSLt0CGZCUTYlc1JrUaNGhw4cIAKFSqwZMkSFEVhzZo1rFy5kgYNGvDf//6XQJkJQIgid+WKhiFDvDCb73YM8fZW2Lo1jbJlC7BjSEoKniNG4DllChqL+vqccdAg0nbvRpFLEMLO8vUdrnbt2uzatYvk5GQuXLiA1WqlWrVqlC9fvrDiE0LkIfNmn94kJqq/n65dm069etYC24/uu+/wGjEC7eXLqnJFo8Ewfz7GsWOx+62yhcDGI7X09HS6devG5s2bAfDz86NJkyY0bdpUEpoQdqIoMH68Jz/9pB5JPXWqge7dzQWzE4sF97AwvDt3zp7QvL1J37IF47hxktBEsWFTUvPy8uKnn37CYinEyU+FEPnyn/+4ERWlngKrc2cTr7+e873J8kvzxx94d++Ox6JF2U43WurXJ/XwYcxduhTIvoQoKDZfU2vVqhVHjx4tzFiEEDY6dMiFuXPVHUNq17awfn16gcwL7LJvH6WefDLHW8VkvPIKqf/7n3QIEcWSzf/+b731FseOHWP27NlcunQJq7XgztcLIWx34YKW4cO9sFrvnvIrXVph69b0fz4FVno6HpMm4T14MNrkZFWVtXx50qKiMCxZAh4euWxACPuyuaNI8+bNs3o8rlmzBq1Wi6urq2oZjUbD1atXCzxIIUSm27cz742WknI3oWm1Cu+/n06NGv/si6b21Cm8Xn4Z3Zkz2epMzzyDfhNJHzQAACAASURBVO1alIoV/9E+hChsNie1nj17opGLwULYjdUKo0d7ceaMumPI3LkGOnT4Bx1D/h575jFrFpoM9fU4xcUFw9y5GEND5X5nwiHYnNTWrl1bKAGEh4ezatUqEhISqFOnDosXL6ZVq1a5Lr9jxw5WrlzJ+fPn8fHxoW3btvz73/+mQoUKhRKfEMXFe++58emn6rMjvXsbGTcu+yw/NjMY8JwwAbcc7rBhqV4d/fvvY2nc+OG3L0QRs+tXr+joaGbMmMHkyZP56quvaN68OX369OH333/PcfnY2FhGjx7NgAED+L//+z+2bNnCmTNnGDlyZBFHLkTRSkrSsHix+jpWw4YWVq3SP3Rvek1CAt7duuWY0IwDBpD65ZeS0ITDyVdS++uvv5gzZw4tW7akUqVKVKpUiZYtWzJv3jz++uuvfO98zZo1DBw4kCFDhlC7dm3CwsKoUKECEREROS7//fffU6lSJUJDQ6lWrRrNmjVj1KhRHDt2LN/7FsKRLF7srrqO5uursGVLGl5eD7c97YkTlGrXDpfvv1eVKz4+pG/YgH7tWvDx+SchC2EXNie1K1eu8NRTT7F69Wo8PT3p1q0b3bp1w8vLi5UrV/LUU09xJZfbt+fEaDRy4sSJbJMgt2vXjm+//TbHdVq0aEFCQgKfffYZiqKQlJREdHQ0zz77rM37FcLRnD6tJSJCPR5t6lQDgYEPNwWWa3Q0pTp3RvvHH6pyS82apB4+jKlPn4eOVQh7s/ma2rx580hJSWHfvn20bt1aVXf06FH69+/P/Pnz2bBhg03bS0pKwmKx4O/vryr39/fnxo0bOa7TvHlz3n//fUaNGoVer8dsNvPMM8888HpfXFycTWXOQtrmmHJqm6LAxInBqu77VaoYaNfuNHFx+UxqViuV1q+nUg5nQlJatuTCokVYMgPJZ+QP5syvGzh3+4pj24KDg3OtszmpHT58mNGjR2dLaJA5MHvUqFG5njYsKGfOnGH69OlMnTqVdu3akZCQwOzZs5kwYQLr16/Pdb37/wBxcXF5/lEcmbTNMeXWtgMHXPj2W29V2ZIlFurVq5m/HaSm4vXKK7h+8km2qowxY2DBAqoX0s08nfl1A+dunyO2zeb/Yr1en+c8j+XLl0ev19u843LlyqHT6UhMTFSVJyYmEhAQkOM6y5cvp0mTJowbNw6A+vXr4+XlRefOnZkzZw6VK1e2ef9CFHdGI8ycqe4c8tRTZrp2zV/3fc1vv+E9YAC606dV5YqrK/rlyzENHvyPYxWiuLD5mlqdOnXYsWMHGRnZ55UzGo1ERUUREhJi847d3Nxo1KgRMTExqvKYmBhatGiR4zp6vR6dTj1G585zmeFEOJvwcDfi4+/+v2u1Cm++mb/ejrojRyjVrl22hGb19ydt3z5JaMLp2HykNmHCBIYOHcozzzzD8OHDqVkz8/RHXFwcH3zwAWfOnOHDDz/M185DQ0MZPXo0jz/+OC1atCAiIoLr168zbNgwAEaPHg2QdWrxueeeY/z48bz//vu0b9+e69ev8/rrr/PYY4/JvdyEU0lK0rB0qfoo7aWXjDRoYPuXN9cPP8Rz8mQ0ZvWRnaVBA9K2bkWR94xwQjYntR49erBu3TrmzJnD1KlTs2YXURSFgIAA1q5dS7du3fK18169enHz5k3CwsJISEggJCSEqKgogoKCALL1phw0aBCpqals2LCBWbNm4evry9NPP828efPytV8hirslS9Rd+H18FGbOtHH2fUXBfckSPJYuzVZl6tGD9HffBW/vHFYUwvFpkpOT89WFymw2c/z48awB0oGBgTRu3BiXQrrIXBgc8eKnraRtjunetv36q5bWrUthsdxNagsW6G2bOURR8JgzB/fVq7NVGWbMIGPatCKf7sqZXzdw7vY5YtvynYlcXFxo1qwZzZo1K4x4hCjRFCWzc8i9Ca1aNQujR9uQ0KxWPKZOxf3999Xb9PIife1azD16FHS4QhQ7Nn9li4yMZHAeF5Vfeukltm7dWiBBCVFSff65C4cPq+d3/Pe/Dbi7P2BFsxnPMWOyJzRfX9L27JGEJkoMm5NaREREnpMGV6xYkfDw8AIJSoiSyGTKuQv/888/oAu/0YjniBHZ5nC0litH6r59WOSsiihBbE5q58+fp169ernWh4SEEB8fXyBBCVES3d+FX6OxoQu/wYDX4MG47d6tKrZWrEja/v1YH3uskKIVoniy+ZqaRqPh5s2budbfvHlTxooJ8ZCSk3UsWXJ/F35T3l34U1PxHjgQl6++UhVbAwNJ27sX66OPFkaoQhRrNh+pPfbYY+zcuTPHwdcGg4GPP/6Yhg0bFmhwQpQUGzZUyqELvyH3FZKT8X7xxWwJzVKjBqmffioJTZRYNie1SZMmcebMGbp06cK+ffuIj48nPj6evXv30qVLF86dO8ekSZMKM1YhnNKvv2rZuVM9NdyUKQYCAnIebaNJSqJUjx643Hc3C0vduqR9+qkMqhYlms2nH5955hneffddpk2bxpAhQ7LKFUXBx8eH1atX06FDh0IJUghnlVsX/ldeybkLv+b6dbx79kT366+qcnOjRqRHR6OULVuo8QpR3OVrnFr//v3p2rUrhw8f5tKlSwBUq1aNdu3a4SM3FBQi3w4etL0Lv+b6dby7dEF34YKq3NyiBWlRUVC6dGGGKoRDyPfgax8fH3rImBch/jGrFebNU3cOad06ly78ZjNew4dnT2ht2pC2datMeyXE32y+pnb27Fn279+vKjty5Ai9evWiffv2vPvuuwUenBDObOdOV06ftq0Lv/uSJbgcPaoqM3XqRNr27ZLQhLiHzUdqs2bNQqPR0LVrVwD++OMP+vXrh7u7O/7+/syaNQs/Pz8GDhxYaMEK4SxMJnjzTfU5xt69TTRsmL0Lv0tMDO7LlqnXb9uW9E2bwM2tUOMUwtHYfKT2008/8eSTT2Y93759O1arlW+++YbY2Fg6deokM4oIYaMtW1y5ePHuUZpOZ+X117MPl9Fcu4bnyJFolLs9Ia0VK6LfsEESmhA5sDmppaSkUK5cuaznBw8e5KmnnuKRRx4BoFOnTjKjiBA20OvJdq+0Hj3+pHr1+47SLBa8Ro5E++efWUWKVkv6hg0o/v5FEaoQDsfmpObv78/ly5cBSE5O5ocffuCZZ57Jqs9pULYQIrvwcDeuXbv71vPwUHj55WvZlnNfuhSXb75RlWVMn47lqacKPUYhHFW+xqm99957+Pr68s3fb7QuXbpk1Z85c4bKlSsXfIRCOJFbt+Cdd9TX0kaONBIQYFKV6b78EvewMFWZuU0bMqZMKfQYhXBkNie1OXPmEB8fz+zZs3Fzc2PBggVZd6g2GAzs3r2bvn37FlqgQjiDNWvcuXnz7lGaj4/CxIkZJCXdXUaTkIDX/dfRAgJI37ABdDqEELmzOan5+/vz2WefkZKSgqenJ273XKRWFIW9e/dSpUqVQglSCGfw558a1qxRH6W99loGZcsqd5PanetoN25kLaNoNJnX0QLUU2kJIbLL9+Dr0jnMWuDp6UmDBg0KJCAhnNU777iTmnp3EFq5clbGjFFfi3YPC8s2SXHGtGlY2rQpkhiFcHQ2dxQRQjy8K1c0hIeru+BPmpTBvbPL6b76CvelS1XLmJ96ioxp04oiRCGcgiQ1IYpAWJgHGRl3j9IqV7by8st3Jy12SUrKfh3N31+uowmRT5LUhChk589r2bxZPWnx9OkGPO4MVbNYqD5nDtqEhKx6RaNB/957KBUrFmGkQji+fF9TE0Lkz5tvuqtuLVOjhoWBA+924XdfvhyP775TrZMxeTLme8aBCiFsI0dqQhSiX37RsnOn+lrazJkZuPz9dVJ39Cjuixer6s2tWpExY0ZRhSiEU8kzqWVkZDBx4kTWrVuX50bWrVvHpEmTMJlMeS4nREmzcKF6OqwGDSy88MLf7xO9Hs/XXkNjvTs9lrV8edLff5+srCeEyJc8k9rGjRvZunUrnTp1ynMjnTp1YsuWLWzatKlAgxPCkcXG6jhwQH0tbfZsA9q/33Xub72V7f5o+vXrUf6eT1UIkX95JrXdu3fz/PPP8+ijj+a5kUcffZTu3bvz8ccfF2hwQjgqRYEFC9RHaS1bmnn22cwbgGp//hn3VatU9Rkvv4y5ffsii1EIZ5RnUjt16hQtW7a0aUPNmzfn1KlTBRKUEI7u8GEXjh5Vn0KcM8eQeQNQsxmvsWPRWCxZdcaAAAxz5xZxlEI4nwdeU/Pw8MhrkSweHh4yU78Q5HyU1qGDiVatMpOY29q16H76SVX/27Rp4OtbZDEK4azyTGoBAQE23yMtPj4ef7nHkxDs2ePCTz+pB0zPmmUAQHvxIh5vvqmqM/bsSYpMgyVEgcgzqbVu3Zrt27eTmpqa50Zu377N9u3beUru8yRKOL0eZs/2VJW98IKRRo2soCh4jh+PRq/PqrP6+WG4b2osIcTDyzOpjRs3jqSkJPr06cOVK1dyXObKlSv069ePpKQkXnvttUIJUghHsWKFO7//fvdtpdMpvPFG5ml51y1bsk1WbFi0SGbfF6IA5TkYJiQkhBUrVjBhwgQaN25Mq1atqFevHqVKlSI1NZXTp09z5MgRFEXhnXfeoW7duvkOIDw8nFWrVpGQkECdOnVYvHgxrVq1ynHZV199lY8++ihbuZeXF1evXs33voUoSJcuaVi5Un1rmVGjjNSqZUWTkIDHrFmqOnObNpgGDizKEIVweg8c4Tlo0CDq1KnDkiVL+PLLL/nqnm+arq6utG3blmnTptGsWbN87zw6OpoZM2awbNkyWrZsSXh4OH369CE2NpbAwMBsyy9ZsoR58+apyjp16pRrEhSiKM2c6YnBcHc6LH9/KzNmZF5L85g+HW1yclad4ulJ+sqVZHaHFEIUFJumLXj88cfZsWMHBoOB8+fPc/v2bXx8fKhevTqenp4P3kAu1qxZw8CBAxkyZAgAYWFhHDp0iIiICObm0L25dOnSqvu5xcbGcunSJdavX//QMQhREP73Pxf271cPtJ4/30Dp0uCyfz9uu3er6gxvvIFSrVoRRihEyZCvuXg8PDyoV69egezYaDRy4sQJxo4dqypv164d3377rU3b+PDDDwkJCaFFixYFEpMQD8NohBkz1F34mzUz07+/CVJS8JwyRVVnbtQI46uvFmWIQpQYeSa1Xbt25Vqn0Whwd3cnMDCQevXqocnnaZSkpCQsFku2YQD+/v7cuOdW9rlJSUlh9+7dzJkz54HLxsXF2VTmLKRtRSsysiLx8XfPIGg0CmPHnuP8+XSCliyh9LVrWXWKTsfZKVPQX7yYbTvFsW0FxZnbBs7dvuLYtuDg4Fzr8kxqw4cPR6PRoNxz48L7aTQaKleuzJIlS+jatevDR5lPUVFRWK1W+vfv/8Bl7/8DxMXF5flHcWTStqJ19aqGiAgfVdnQoUa6d6+M7uhRSu3cqarLGD+eKs8/n207xbFtBcWZ2wbO3T5HbFueSW3fvn15rqzX6zl79ixbtmxhyJAh7NmzhyeffNKmHZcrVw6dTkdiYqKqPDExkQAbujh/+OGHdO/enTJlyti0PyEKw5w5HqSl3T1L4ednZfbsDDAY8Bw/XrWspUYNMqZOLeoQhShR8kxqrVu3fuAGnn32WYYNG0abNm1YsWKFzUnNzc2NRo0aERMTwwsvvJBVHhMTQ/fu3fNc99ixY5w8eZLF992HSoiidOSIjo8/Vt8rbfbsDMqWVXBf+Da6+07b6FeuhH/QsUoI8WAFcpNQb29v+vTpww8//JCv9UJDQ9m6dSuRkZGcPXuW6dOnc/36dYYNGwbA6NGjGT16dLb1Nm7cSI0aNWQGE2E3ZjNMnapOUA0aWBg61Ij25EncV6xQ1WUMHYrFhi+JQoh/psDuRFihQgXS09PztU6vXr24efMmYWFhJCQkEBISQlRUFEFBQQA5zmJy+/ZtoqOjmTZtWoHELcTDeP99N06fVs/vGBamR2cy4DVqFBqzOavcWrEihvvGVwohCkeBJbUzZ87YdC3sfiNGjGDEiBE51u3fvz9bmY+PD3/88Ue+9yNEQUlM1PDmm+ou/P36GWnZ0oLH9LnoTp9W1enDwsDPryhDFKLEKpDTj8ePHycyMpJnn322IDYnRLG2YIEHKSl3O4f4+CjMn2/A5eBB3O+bCMDYty/mbt2KOkQhSqw8j9QmT56c58p6vZ74+Hh+/PFHKlSowPTp0ws0OCGKm2PHdGzapO4cMm2agUe0CXiOGaMqtwYFZR6lCSGKTJ5JLSIiIs+V7wy+HjVqFBMnTpT7qQmnZrXC1Knq0461a1t4ZXQGnv96De09w1MUnY70DRvgnmndhBCFL8+k9tdffxVVHEIUe5s3u/Ljj+q3zNKlerw/eA/Xzz9XlWdMnYpFpm8TosgVyDW1Oy5cuFCQmxOi2EhOhvnz1Udp3bubaOf/Mx73TdVmbtmSjPvmexRCFI1/nNSSkpJ477336NChA02bNi2ImIQodhYu9CAp6e7bxdNTYdGsm3iNHIkmIyOrXPH1JX39enApsI7FQoh8eKh3nl6vZ//+/URFRfHFF19gMpmoUaOG3PlaOKUfftDx/vvqziGTJmUQHD4ne/f9ZctQqlYtyvCEEPewOakpikJMTAzbt2/n008/JTU1FY1Gw+DBg3nttdccbtJLIWxhMsH48Z4oyt0u/NWrW5gcsg/3Re+pljX264epT5+iDlEIcY8HJrUTJ06wfft2du3aRUJCAjVq1GDMmDE0adKE/v370759e0lowmmtW+fGqVPqmUPWzPmN0hNDVWWWatWk+74QxUCeSa158+bEx8dTqVIl+vTpw4svvkijRo0AuJjD/aCEcCa//aZh8eL7Zg7pY6D95tHZuu/rN2wAX9+iDlEIcZ88k1pcXBxVq1Zl3rx5dO7cGXd396KKSwi7UpTMCYvT09W3lVldayWui/6nWjZj+nQszZoVdYhCiBzk2ftx1apVBAUF8fLLLxMcHMzo0aM5ePAgFoulqOITwi727nXh889dVWVrRn9P+bD7uu8/8QQZD5h5RwhRdPI8Uhs8eDCDBw/m6tWr7Nixg6ioKKKioihbtixPPvkkGo0GjUaT1yaEcDgpKTB9uvq2Mm2a36bf7qFojMassqzu+zrd/ZsQQtiJTePUKlWqxPjx4zly5Ahff/01gwYN4scff0RRFCZNmkRoaCiffPIJaWlphR2vEIVu4UIPrl+/+9ZwdVXYHjgJ3dkzquX077yD8vdtkoQQxUO+x6nVr1+f+vXrM3/+fL7++mu2b9/Ovn372Lp1Kx4eHly7dq0w4hSiSPzwg47wcPWYtA2dt1NhZ7iqzNi/P6YXXyzK0IQQNnjoGUU0Gg1PP/00a9asIS4ujoiICNq2bVuAoQlRtHIak9Y68AKDv3xFtZylWjX0b71V1OEJIWxQIHP5uLu707NnT3r27FkQmxPCLu4fk+aCiT1eA9D+npJVpri6ov/gA+m+L0QxVaATGgvhqHIak7Y9ZDZlz36nKjPMnYulceOiDE0IkQ+S1ESJl9OYtB7en9PzjPoUo6ljR4z33QhUCFG8SFITJd79Y9ICSGCL7iU0ipJVZq1YEf2774JW3jJCFGfyDhUl2v1j0jRY2VP6JbxvJWSVKRoN6e+9h1K+vD1CFELkgyQ1UaLdPyZtmvZtWqbcdxfrKVOwPP10UYcmhHgIktREiXX/mLTmfMsiZqqWMT/xBBnTpxd1aEKIhyRJTZRIVitMmeKRNSatNMl87NIfndV8dxk/P9I3bJC7WAvhQCSpiRJpyxZXTpy4k6wU3mMUgeZLqmX0a9agVKlS5LEJIR6eJDVR4qSkwIIFd8ekjSCcvuxQLZMxahTmrl2LOjQhxD8kSU2UOGFhHiQmZv7r1+UUqxinqrc0aIBhwQJ7hCaE+IckqYkSJS5Oy7p1mZ1DPElnO/3wxJBVr3h7kx4RAR4euW1CCFGMSVITJcrMmR6YzZmdQ5Ywg/qcUtXr334ba3CwPUITQhQASWqixPj887szh7Tk/3iN/6jqjf36YRowwB6hCSEKiCQ1USIYjfDGG5mnFF0xEs4ItNydBstSrRr6t9+2V3hCiAJi96QWHh5Ow4YNqVChAm3atOHo0aN5Lm80Glm0aBENGzYkICCA+vXrs27duiKKVjiq9evdiI/PvK3MDJZQj9Oqev2KFeDjY4/QhBAFyK6jSqOjo5kxYwbLli2jZcuWhIeH06dPH2JjYwkMDMxxneHDh3P16lVWrlxJ9erVSUxMRK/XF3HkwpHcuKEhLCzzKK0OvzKTRap648CBWOQGt0I4BbsmtTVr1jBw4ECGDBkCQFhYGIcOHSIiIoK5c+dmW/7w4cN89dVXHD9+nHLlygFQtWrVIo1ZOJ4FCzy4dUuDBisbGIk7xqw6q78/hoUL7RidEKIg2e30o9Fo5MSJE7Rr105V3q5dO7799tsc19m/fz+NGzdmzZo11K1blyZNmjBt2jRSU1OLImThgI4f17FlS2bnkNGspzVHVPWGpUtRypa1R2hCiEJgtyO1pKQkLBYL/v7+qnJ/f39u3LiR4zqXLl0iNjYWd3d3IiMjSUlJYdq0aVy/fp3IyMhc9xUXF2dTmbOQtmVSFBg/vg6KoqEyV1iKemLi5Natia9fH4rJ30teN8flzO0rjm0LzmPYjUPN1Gq1WtFoNGzYsIHSpUsDmacse/XqxY0bNwgICMhxvfv/AHFxcXn+URyZtO2uqChXfv7ZC1B4lzH4cjurTilVCu26dQQXk7kd5XVzXM7cPkdsm91OP5YrVw6dTkdiYqKqPDExMdfkVKFCBR555JGshAZQq1YtAK5cuVJ4wQqHk5oKc+dmdg7pzcd0Z5+q3jB3rkxWLIQTsltSc3Nzo1GjRsTExKjKY2JiaNGiRY7rtGzZkuvXr6uuoZ0/fx4g196SomRascKda9e0lOEm/+E1VZ25eXOML79sp8iEEIXJruPUQkND2bp1K5GRkZw9e5bp06dz/fp1hg0bBsDo0aMZPXp01vK9e/embNmyhIaG8uuvvxIbG8uMGTPo0aNHtmtzouS6dEnD6tXuAIQxlQrcvUaruLqiX7UKtHYfoimEKAR2vabWq1cvbt68SVhYGAkJCYSEhBAVFUVQUBCQ/ZRiqVKl2L17N9OmTaNdu3b4+fnRtWvXHLv/i5Jr9mxPMjI0PMNhXiZCVZcxaRLWOnXsFJkQorDZvaPIiBEjGDFiRI51+/fvz1YWHBzMrl27Cjss4aBiYlzYt88VD/S8xyhVnaV2bTImTbJTZEKIoiDnYITTuHhRy4gRngDMZT41OZ9Vp2g0macd3d3tFZ4QoghIUhNOITkZ+vb1IilJSyOOMwX15MTGESOw5NIBSQjhPCSpCYdnNMLgwd7ExenQYSacEbhgyaq3Vq6MYfZsO0YohCgqktSEQ1MUmDjRk6+/zrw8PI95PM6PqmX0y5aBr689whNCFDFJasKhvfOOO1u2uAEwllXMun8G/l69MD/3nD1CE0LYgSQ14bB273ZhwYLMWUMGsZlVjFfVW8uWxbBkiT1CE0LYiSQ14ZC+/17HK694AdCF/WxkqKpe8fIifds2lFymXBNCOCdJasLh/PabhoEDvTAYNLTmaz6mt6pjiOLqSvqmTViaN7djlEIIe5CkJhxKcjL06+dNYqKWxzjBJzyPJ4asekWjQb9+Peb27e0YpRDCXiSpCYdhNmsYOtSLM2d01CCe//IcpbmlWsawbBmmXr3sFKEQwt4kqQmHoCiwdGkQX3zhyiNc5SDPUpEE1TKGWbMwDh9upwiFEMWB3ed+FMIWq1e7sXt3acpwk8/pyKNcUtVnjBlDxuTJ9glOCFFsyJGaKPb273dh7lwPvEhjP12pzylVvXHAAAwLF4JGY6cIhRDFhSQ1UaydO6fllVe8cFFMRNOLJ4hV1Zs6d0a/erXcH00IAcjpR1GM3boF//qXF2m3rWzhJTrxuare3KoV6RER4CL/xkKITPJpIIolqxXGjPHi3DkdKxhPf7ar6i0NG5L20Ufg6WmnCIUQxZGcsxHF0ooV7nzyiSuh/IfxrFLVWWrUIG3nTihd2k7RCSGKK0lqotg5fNiFf//bnc58ysr753N85BHSdu1C8fe3U3RCiOJMTj+KYuXSJQ3Dh3tSX/mF7fRDhzWrzuLpSfr27ShBQXaMUAhRnMmRmig20tMzb/bpkZzAJzyPD6lZdYpWy4U338TasKEdIxRCFHeS1ESxoCgwYYIn8b9ksJfuBPG7qt7w5puktG5tp+iEEI5CkpooFt57z40dUS5sYjDN+EFVlzFyJMbRo+0UmRDCkcg1NWF3R4/qmDnTgzd5nReJVtWZOnTAsHixzBYihLCJHKkJu7p6NXPm/SHm95nBUlWdpW5dGVwthMgXSWrCbjIyYMgQL+rfiGEtr6rqrAEBpG3fDr6+dopOCOGI5CuwsJvXX/fg9vdxfM6LuGLOKlc8PUnftg0lMNCO0QkhHJEkNVHkvv9ex/Ll7nz/WTLf0hU/UlT16evWYWnSxE7RCSEcmSQ1USQUJXOmkOXL3TlyxIVHuMpu+lCdi6rl9PPmYe7Rw05RCiEcnSQ1UagsFtizx5V33nHn7C9murOL/XxAJw6oZgsBMA4ejHH8+Fy2JIQQDyZJTRQKgwG2bXNl5Up3Sl/8mZFEMJCtlONmjsubn34a/bJl0nVfCPGPSFITBerWLdi40Y1tq2/xbOIHRLORRvyU5zqWevVIi4wEN7ciilII4azsntTCw8NZtWoVCQkJ1KlTh8WLF9OqVascl/3666/p1q1btvLvvvuOWrVqFXao4j5GI5w+reXECR0nvzeR/P0FPM7/yguWnRxnH26Y8lzfUr8+xkGDMA4dKvdFE0IUCLsmtejoaGbMmMGyZcto2bIl4eHh9OnTh9jYWALz6M4dGxtLmTJlsp6XL1++NxrkGAAAEBZJREFUKMIt0Uwm+PW0hrPf3OTPb+Ixn4zD549zBFvP0o0zjOUSWpQHbsdapgymPn0wDhqE9bHHiiByIURJYtektmbNGgYOHMiQIUMACAsL49ChQ0RERDB37txc1/P396dcuXJFEuNPiw5m3obZQdy8+RepfhdQlMwehwCKosGqAH+X3XlYM0xY0gxYUw1Y0/Qo6QbQG8CgR2swoM3QozMa0Jn0lE6/Tj3lDE/xV75jUrRazB06YBw0CPNzz4G7e8E2Wggh/ma3pGY0Gjlx4gRjx45Vlbdr145vv/02z3Xbtm2L0Wikdu3aTJkyhaeffrrQ4mwWNhhPDIW2fWdmqh6M+aVBmPr1Q3nkEXuHI4QoAeyW1JKSkrBYLPjfdwdjf39/bty4keM6FStWZPny5TRp0gSj0cj27dvp0aMH+/fvz/U6nCgaVjToA4IgpBYuDetg6tYNS7Nm0ptRCFGk7N5RJD+Cg4MJDg7Oet68eXMuX77MqlWr8kxqcXFxNpXlpF7+w3RqBq0nN8pURx9UDV29QDQhgRiqVSMjKAirh4d64fj4At+/ra+bI5K2OS5nbl9xbNu9eeB+dktq5cqVQ6fTkZiYqCpPTEwkICDA5u08/vjjREdH57nM/X+AuLi4PP8o9zr+SGe0ZqPN8dibxWLBxUWX+UQDmnt/cs+BkwYUrQ6ruwd4eKJ4eqDx9ETj5YG2lAc6b09cfDxwKeWOq68Hno/44tO0JkqVKpTWaild9E3L1+vmaKRtjsuZ2+eIbbNbUnNzc6NRo0bExMTwwgsvZJXHxMTQvXt3m7fzyy+/UKFChcIIEYB6v35QaNsuDIX5T/jgvo1CCGFfdj39GBoayujRo3n88cdp0aIFERERXL9+nWHDhgEw+u+7Ha9fvx6Ad999l6CgIEJCQjAajURFRbF//34iIyPt1gYhhBDFh12TWq9evbh58yZhYWEkJCQQEhJCVFQUQUFBAFy5ckW1vMlkYs6cOVy9ehUPD4+s5Tt27GiP8IUQQhQzdu8oMmLECEaMGJFj3f79+1XPx48fz3iZ8FYIIUQu5M7XQgghnIYkNSGEEE5DkpoQQginoUlOTpae2kIIIZyCHKkJIYRwGpLUhBBCOA1JakIIIZyGJDUhhBBOQ5KaEEIIp1Hiklp4eDgNGzakQoUKtGnThqNHj9o7pH9s8eLF+Pn5qR61atWyd1gP5ciRI/Tv35+QkBD8/PzYsmWLql5RFBYvXkydOnWoWLEiXbt25ddff7VTtPnzoLa9+uqr2V7HDh062Cna/Fm+fDnPPPMMgYGB1KhRg379+nH69GnVMo762tnSNkd97TZs2ECrVq0IDAwkMDCQZ599lgMHDmTVO+JrVqKSWnR0NDNmzGDy5Ml89dVXNG/enD59+vD777/bO7R/LDg4mLNnz2Y9HDVZp6WlUbduXZYsWYKnp2e2+pUrV7JmzRqWLl3K4cOH8ff3p2fPnty+fdsO0ebPg9oGmXd1v/d13LFjRxFH+XC++eYbXn75ZQ4cOMDevXtxcXHhhRde4K+//spaxlFfO1vaBo752lWqVIn58+fz5ZdfEhMTw9NPP82gQYM4efIk4JivWYkap9a+fXvq1avHqlWrssqaNGlCjx49mDt3rh0j+2cWL17M3r17+b//+z97h1KgKleuzFtvvcWgQYOAzG+NderUYeTIkUyZMgUAvV5PcHAw//73v7Pu7uAI7m8bZH7bv3nzJtu3b7djZAUjNTWVoP9v7/5joi7jAI6/8RQFAQ/lOiBAQGkk0SiIgpYSBambMmVKYEx21YiZ8I+/iIorrNwOE5Ef0mKxBHFOtpJWYT/QsQlZQ7PVWikBqQyUguCSIXf2B+PmHT+CQs87Pq/Nze/jc9/n8/Bx9+H7fH/5+VFVVcWqVavsKneWcwP7yp2/vz+5ubmkpaXZZM5mzJHa4OAg586dIzY21qw9NjaWb775xkpRTZ/W1laCg4N58MEH0Wg0tLa2WjukadfW1kZnZ6dZDp2cnIiOjraLHAI0NjaydOlSwsPDyczMHPUSXVvR39+P0WhEqVQC9pU7y7mNsPXcGQwGampq0Ov1REZG2mzOrP6U/julu7sbg8GASqUya1epVHR1dVkpqukRERFBSUkJQUFBXLt2DZ1OR3x8PE1NTSxcuNDa4U2bzs5OgDFz2NHRYY2QptXTTz/NmjVrWLx4Me3t7ezevZu1a9dy8uRJ5s6da+3wpmTXrl2EhoYSGRkJ2FfuLOcGtp27H3/8kfj4eAYGBpg/fz6VlZWEhISYCpet5WzGFDV7FhcXZ7YdERFBWFgYhw8f5uWXX7ZSVGKqEhMTTX8PCQkhLCyM0NBQ6urqpvQ2eGt75ZVXaGpq4vPPP0ehUFg7nGk13txsOXdBQUE0NDTw119/8fHHH5ORkcEnn3xi7bD+sxmz/Lho0SIUCsWoJYGrV69yzz33WCmq28PFxYXg4GBaWlqsHcq0UqvVADMihwBeXl54e3vbVB6zs7Opqanh+PHj+Pv7m9rtIXfjzW0stpQ7R0dHAgMDCQsLIzc3l9DQUEpKSmw2ZzOmqDk6OhIWFkZ9fb1Ze319PY8++qiVoro9BgYG+PXXX03/Ke3F4sWLUavVZjkcGBigsbHR7nIIw0vmHR0dNpPHnTt3mr70LW8psfXcTTS3sdha7m5lNBoZHBy02Zwpdu3apbV2EHeKq6sr77zzDp6ensybNw+dTsfp06cpKipiwYIF1g7vP3v11VdxdHTEaDRy4cIFtm/fTktLC/v27bO5efX39/Pzzz/T2dnJoUOHWLZsGW5ubgwODrJgwQIMBgMFBQUsWbIEg8FATk4OnZ2dFBQU3PXnLiaam0Kh4M0338TFxYWhoSF++OEHtm7disFgQKfT3fVz27ZtG0eOHKGiogIfHx/0ej16vR4Y/oXSwcHBZnP3b3Pr7++32dxptVrTd8fly5cpLS3l6NGjaLVaU55sLWcz6pJ+GL75ev/+/XR2dnL//ffz9ttv8/jjj1s7rP9Fo9Fw+vRpuru78fDwICIigpycHIKDg60d2pQ1NDSwZs2aUe3JycmUlpZy8+ZN9uzZQ0VFBT09PYSHh5Ofn8+yZcusEO3UTDS3d999l02bNnH+/Hl6e3tRq9U88cQT5OTk4OPjY4Vop8bySsARO3fuJDs7G8Bmc/dvc7t+/brN5i4jI4OGhga6urpwc3MjJCSEzMxMnnrqKcA2czbjipoQQgj7NWPOqQkhhLB/UtSEEELYDSlqQggh7IYUNSGEEHZDipoQQgi7IUVNCCGE3ZCiJoSd+vLLL1EqlXz77bdWHd+WnyMobI880FgIC+PdbGupuLjY7H1oM9XBgwdxd3cnKSnJ2qEIIUVNCEtlZWVm2xUVFXz33XcUFRWZtd/Nz7+7k8rKyggMDJSiJu4KUtSEsGD55Xzy5Emam5sn/aU9NDSE0WjE0dHxdoQnhJiAnFMT4n/45ZdfUCqVlJSUUFxcTFhYGGq1mu+//x4YfuJ5UVERjz32GGq1mqCgILKysujp6Rm1r7q6OlauXIm3tzc+Pj5s3LiRn376aVJxtLW1kZSUhLe3N0FBQbz22mvcuHFjzL5nzpxh/fr1+Pn54eXlxerVq2lsbDTro9VqUSqVXLx4kbS0NHx9fQkICGDbtm38/fffpn733Xcfv/32G1999RVKpRKlUskjjzxiti+j0ciePXsIDg7G09OTdevW0dbWNql5CTFVcqQmxDT48MMPGRgYYPPmzTg5OeHh4QHAli1bOHbsGJs2bSI9PZ329nbee+89zp07xxdffGE6mqusrGTr1q3ExcWh1WoZGBigvLyclStXcurUKQICAsYdu6+vj7Vr13L16lXS09NRqVQcOXKEr7/+elTf+vp6Nm7cSHh4ODt27GDWrFlUV1eTkJBAbW3tqCXV1NRUfH19ef311zl79izvv/8+HR0dVFVVAaDT6di+fTsqlYrMzEwA3NzczPYx8qT6rKws/vjjDw4cOEBGRgaffvrpf/+BCzEOKWpCTIMrV67Q3NxsKmYAp06dorq6mg8++IB169aZ2mNiYkhISKCmpobk5GR6e3vJzs5Go9Gwd+9eU7+UlBQiIiLIz8+nuLh43LHLy8tpa2vj8OHDrF69GoDNmzcTHR1t1s9gMJCVlcWTTz7J0aNHTe0ajYbo6Gh2795NbW2t2Wf8/f2pqqrCwcEBAA8PDwoLC2lsbCQqKoqEhAS0Wi2enp7jLs8ajUY+++wz5syZAwy/xDY3N5eWlhYCAwMn/LkKMVWy/CjENEhISDAraAAfffQR7u7uLF++nO7ubtOfBx54ADc3NxoaGoDhS9/7+vpITEw06+fg4EBkZKSp33hOnDiBj4+PqaABzJ8/n+eee86sX3NzM+3t7WzYsMFsHL1eT0xMDE1NTaOWLNPT000FbWQbhpdKJyslJcVU0ACioqIAaG1tnfQ+hJgsOVITYhqMtTx44cIF/vzzT5YsWTLmZ65du2bqB5gVpVs5OztPOPbvv/8+5hHP0qVLR8UD8OKLL467r97eXrPibLnfe++9FycnJ9rb2yeM6VaW7xQbuWVirPOKQvxfUtSEmAbz5s0b1WY0GvH09OTgwYNjfmbhwoWmfjD8AlvLoz2AWbOmZ0FlZJy33nqLkJCQMftYng+bDgqFYsz2mzflVY5i+klRE+I2CQgI4MyZM0RFRTF37twJ+wGoVCpWrFgx5XF8fX1paWkZ1T5yZGY5jpubGzExMZPad0tLC76+vqbty5cvc/36dfz8/Extty5PCmFtck5NiNtk/fr13Lhxg/z8/FH/NjQ0ZFp+e+aZZ3BxcWHv3r0MDQ2N6juyTDme+Ph4Ll26ZHY1oV6vp7Ky0qxfZGQkPj4+HDhwAL1eP6lxLG9EH9mOj483tTk7O8tSorhryJGaELdJbGwsqamp6HQ6zp8/z4oVK5gzZw4XL17k+PHj5OXlkZiYiLu7Ozqdji1btrB8+XISExNZtGgRly5d4sSJE4SHh7Nv375xx3n++ecpLy/nhRde4KWXXkKlUlFdXY2Li4tZv9mzZ1NUVERSUhJRUVGkpKTg5eXFlStXaGhowNnZmWPHjpl9prW1laSkJOLi4jh79ixVVVWsWrXK7MrKhx56iKqqKnQ6HYGBgbi6upoVPSHuJClqQtxGhYWFPPzww1RUVJCXl8fs2bPx9fVlw4YNpqsAAZKTk/H29qagoID9+/czODiIl5cXUVFRpKamTjiGq6srtbW17Nixg9LSUlxcXHj22WeJjo4mOTnZrG9MTAx1dXXk5+dTVlaGXq9HrVYTHh5OWlraqH0fOnSIvLw83njjDRQKBRqNhry8PLM+OTk5dHV1UVhYSF9fH0FBQVLUhNU49PT0yNlaIYQZrVZLQUEBra2tk37AsxB3AzmnJoQQwm5IURNCCGE3pKgJIYSwG3JOTQghhN2QIzUhhBB2Q4qaEEIIuyFFTQghhN2QoiaEEMJuSFETQghhN6SoCSGEsBv/AAF8oS1Sih9NAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "max_depths = np.linspace(1, 32, 32)\n",
    "train_results = []\n",
    "test_results = []\n",
    "for max_depth in max_depths:\n",
    "   rf = RandomForestClassifier(max_depth=max_depth, n_estimators=300,  n_jobs=-1)\n",
    "   rf.fit(X_train, y_train)\n",
    "   train_pred = rf.predict(X_train)\n",
    "   false_positive_rate, true_positive_rate, thresholds = roc_curve(y_train, train_pred)\n",
    "   roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "   train_results.append(roc_auc)\n",
    "   y_pred = rf.predict(X_test)\n",
    "   false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, y_pred)\n",
    "   roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "   test_results.append(roc_auc)\n",
    "from matplotlib.legend_handler import HandlerLine2D\n",
    "line1, = plt.plot(max_depths, train_results, 'b', label=\"Train AUC\")\n",
    "line2, = plt.plot(max_depths, test_results, 'r', label=\"Test AUC\")\n",
    "plt.legend(handler_map={line1: HandlerLine2D(numpoints=2)})\n",
    "plt.ylabel('AUC score')\n",
    "plt.xlabel('Tree depth')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Increasing the number of estimators increases the model size greatly, but soon additional estimators add very little predictive power"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9644452548090072\n",
      "-rw-r--r-- 1 todd todd 626M Mar 29 22:55 /tmp/tmp.300.estimator.tree.pkl\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(rf.score(X_test, y_test))\n",
    "dump(rf, '/tmp/tmp.300.estimator.tree.pkl')\n",
    "! ls -lh /tmp/tmp*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9646275868356277\n",
      "-rw-r--r-- 1 todd todd 626M Mar 29 22:55 /tmp/tmp.300.estimator.tree.pkl\n",
      "-rw-r--r-- 1 todd todd 1.7G Mar 29 22:55 /tmp/tmp.800.estimator.tree.pkl\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=800, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(rf.score(X_test, y_test))\n",
    "dump(rf, '/tmp/tmp.800.estimator.tree.pkl')\n",
    "! ls -lh /tmp/tmp*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run Grid Search to optimize one of the best classifiers\n",
    "often one runs GridSearch to determine the parameters of the best classifier. However, it is fast a reliable to narrow the parameter window one at a time. Again, the general guidelines:\n",
    "* scoring 100% on the training set indicates overfitting, and a decreased ability to generalize\n",
    "* the difference between the test and validation sets indicate how well the data has been split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 1.0\n",
      "Test score: 0.9644452548090072\n",
      "Validation score: 0.9616494970674933\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.9947773877822131\n",
      "Test score: 0.9580940225483939\n",
      "Validation score: 0.9575470264685325\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, max_depth=25, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 1.0\n",
      "Test score: 0.970249490989759\n",
      "Validation score: 0.9678487859725894\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, max_features=0.4, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 1.0\n",
      "Test score: 0.9644452548090072\n",
      "Validation score: 0.9616494970674933\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, max_features='auto', n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 1.0\n",
      "Test score: 0.9644452548090072\n",
      "Validation score: 0.9616494970674933\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, max_features='sqrt', n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.989789207035549\n",
      "Test score: 0.9552070987935698\n",
      "Validation score: 0.9540827179627435\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, min_samples_leaf=2, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.9689443420615643\n",
      "Test score: 0.9436897924453763\n",
      "Validation score: 0.942261524903516\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, min_samples_leaf=4, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.9959104732259724\n",
      "Test score: 0.9614367763697693\n",
      "Validation score: 0.960160452183426\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, min_samples_split=5, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.9861945911450024\n",
      "Test score: 0.9560275929133619\n",
      "Validation score: 0.9552070987935698\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300, min_samples_split=10, n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.989789207035549\n",
      "Test score: 0.9552070987935698\n",
      "Validation score: 0.9540827179627435\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300,\n",
    "                            min_samples_split=4, min_samples_leaf=2,\n",
    "                            n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training score: 0.9943801566783665\n",
      "Test score: 0.9667244051417632\n",
      "Validation score: 0.9646883641778345\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(n_estimators=300,\n",
    "                            min_samples_split=4, min_samples_leaf=2,\n",
    "                            max_features=0.4,\n",
    "                            n_jobs=-1, random_state=12)\n",
    "rf.fit(X_train, y_train)\n",
    "print(f\"Training score: {rf.score(X_train, y_train)}\")\n",
    "print(f\"Test score: {rf.score(X_test, y_test)}\")\n",
    "print(f\"Validation score: {rf.score(X_validation, y_validation)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Comments on the model parameters\n",
    "* `min_samples_split` (default = 2) The minimum number of samples required to split an internal node\n",
    "    * increasing this number has the effect of nudging the model towards generalization; it takes several examples to fork a branch in the decision tree\n",
    "* `min_samples_leaf` (default=1) The minimum number of samples required to be at a leaf node.\n",
    "    * \"This may have the effect of smoothing the model, especially in regression.\" Increasing this number may encourage generalization, as a single example may not trigger a decision point/tree branch.\n",
    "\n",
    "### We will still use grid search because:\n",
    "* it will give us easy cross validation of our scores\n",
    "* it will provide us with an easy way to extract our parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best score: %s 0.9709996945896789\n",
      "Best params %s {'max_features': 0.4, 'n_estimators': 300}\n"
     ]
    }
   ],
   "source": [
    "grids = GridSearchCV(cv=5, error_score='raise',\n",
    "                     estimator=RandomForestClassifier()\n",
    "                     , n_jobs=-1,\n",
    "                     param_grid={\n",
    "                        'n_estimators' : [300],                   \n",
    "                        #  'max_depth': [25],\n",
    "                        'max_features': [0.4], # 'auto', 'sqrt', 0.4],\n",
    "                        'min_samples_leaf': [2] ,#1, 2, 4],\n",
    "                        'min_samples_split': [4], #2, 5, 10]\n",
    "                        })\n",
    "grids.fit(all_X, all_y)\n",
    "print('Best score: %s', grids.best_score_)\n",
    "print('Best params %s', grids.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using the best parameters from GridSearch, build the optimal classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "                       max_depth=None, max_features=0.4, max_leaf_nodes=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, n_estimators=300,\n",
       "                       n_jobs=None, oob_score=False, random_state=None,\n",
       "                       verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# First let's copy the parameters for the provenance file\n",
    "mdl_params = deepcopy(grids.best_params_)\n",
    "# Let's also remove the base_estimator parameters, since they aren't honored by the constructor, unlike GridSearch\n",
    "if 'base_estimator__criterion' in mdl_params:\n",
    "    del mdl_params['base_estimator__criterion']\n",
    "classifier = RandomForestClassifier(**mdl_params)\n",
    "classifier.fit(all_X, all_y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save the classifier, so it can be loaded without training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['is_transliterated_greek.mdl.0.21.3.joblib']"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_output_file = 'is_transliterated_greek.mdl.{}.joblib'.format(sklearn.__version__)\n",
    "dump(classifier, model_output_file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save model provenance \n",
    "#### Provenance records:\n",
    "* History of the classifier\n",
    "* What data was used\n",
    "* Which hyperparameters had what values\n",
    "* What algorithm was used\n",
    "* What score was achieved\n",
    "\n",
    "#### Important for allowing others:\n",
    "* To use the classifier in the future without rebuilding from scratch\n",
    "* To determining how it may be retrained for better performance\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wrote provenance file: is_transliterated_greek.mdl.0.21.3.joblib.prov.json\n"
     ]
    }
   ],
   "source": [
    "data_files = {}\n",
    "idx = 1\n",
    "\n",
    "for idx, file in enumerate(good_files, 1):\n",
    "    data_files[idx] = {\"filename\": file[file.rfind(\"/\") + 1:],\n",
    "                       \"md5\": md5(os.path.join(latin_reader.root, file))\n",
    "                       }\n",
    "\n",
    "for idx, file in enumerate(greek_texts, len(good_files)):\n",
    "    data_files[idx + 1] = {\n",
    "        \"filename\": file[file.rfind(\"/\") + 1:],\n",
    "        \"md5\": md5(os.path.join(perseus_greek.root, file))\n",
    "    }\n",
    "\n",
    "provenance_file = '{}.prov.json'.format(model_output_file)\n",
    "\n",
    "params = {\n",
    "    \"provenance_data\": provenance_file,\n",
    "    \"date_created\": str(datetime.datetime.now()),\n",
    "    \"model_parameters\": mdl_params,\n",
    "    \"max_word_length\": max_len,\n",
    "    \"num_samples\": num_samples,\n",
    "    \"num_features\": num_features,\n",
    "    \"library_version\": sklearn.__version__,\n",
    "    \"classifier_class\": \"{}\".format(str(classifier.__class__)),\n",
    "    \"classifier_best_score\": grids.best_score_,\n",
    "    \"data_files\": data_files,\n",
    "    \"model_output_file\": model_output_file,\n",
    "    \"model_output_md5\": md5(model_output_file),\n",
    "    \"labels\": label_encoder.classes_.tolist(),\n",
    "    \"best_score\": grids.best_score_,\n",
    "    \"best_params\": grids.best_params_,\n",
    "    # manually added information\n",
    "    \"comment\": \"Transliterated Greek Classifier\",\n",
    "    \"code_generated_by\": \"loanwords_problems_solutions.ipynb\",\n",
    "    \"feature_encoding_fun\": \"encode_word\",\n",
    "    \"char_encoding_file\": \"latin.greek.char_idx.pkl\",\n",
    "    \"author\": \"Todd Cook\",\n",
    "    \"sample_pipeline\": process_latin_text_pipeline_file\n",
    "}\n",
    "\n",
    "with open(provenance_file, 'wt') as writer:\n",
    "    json.dump(params, writer, indent=2)\n",
    "    print('Wrote provenance file: {}'.format(provenance_file))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### How about that model provenance file? It should be readable, here it is:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"provenance_data\": \"is_transliterated_greek.mdl.0.21.3.joblib.prov.json\",\n",
      "  \"date_created\": \"2020-03-30 09:09:33.988893\",\n",
      "  \"model_parameters\": {\n",
      "    \"max_features\": 0.4,\n",
      "    \"n_estimators\": 300\n",
      "  },\n",
      "  \"max_word_length\": 28,\n",
      "  \"num_samples\": 219377,\n",
      "  \"num_features\": 28,\n",
      "  \"library_version\": \"0.21.3\",\n",
      "  \"classifier_class\": \"<class 'sklearn.ensemble.forest.RandomForestClassifier'>\",\n",
      "  \"classifier_best_score\": 0.9709996945896789,\n",
      "  \"data_files\": {\n",
      "    \"1\": {\n",
      "      \"filename\": \"bc1.txt\",\n",
      "      \"md5\": \"3bbb7a0755de2c548bd42a286374d809\"\n",
      "    },\n",
      "    \"2\": {\n",
      "      \"filename\": \"bc2.txt\",\n",
      "      \"md5\": \"1d41b8371fb64407bf2701fc2567d8c3\"\n",
      "    },\n",
      "    \"3\": {\n",
      "      \"filename\": \"bc3.txt\",\n",
      "      \"md5\": \"43ba4f905803f74a6c7c79ceec5b991f\"\n",
      "    },\n",
      "    \"4\": {\n",
      "      \"filename\": \"bellafr.txt\",\n",
      "      \"md5\": \"e42fb6ac2b7b5060bad4d86cbdef6d53\"\n",
      "    },\n",
      "    \"5\": {\n",
      "      \"filename\": \"gall1.txt\",\n",
      "      \"md5\": \"1b4bf3b593e6a7d33dadc518719e3e8d\"\n",
      "    },\n",
      "    \"6\": {\n",
      "      \"filename\": \"gall2.txt\",\n",
      "      \"md5\": \"ad7deff6f37a869c095b66ec5f974fec\"\n",
      "    },\n",
      "    \"7\": {\n",
      "      \"filename\": \"gall3.txt\",\n",
      "      \"md5\": \"a68dadc4d308e1e47ca714a00b9db36c\"\n",
      "    },\n",
      "    \"8\": {\n",
      "      \"filename\": \"gall4.txt\",\n",
      "      \"md5\": \"5d59a6fc9d6d66c7f505d208e9b90eaa\"\n",
      "    },\n",
      "    \"9\": {\n",
      "      \"filename\": \"gall5.txt\",\n",
      "      \"md5\": \"875b11f7ba474f2e682ebf57cdf2af91\"\n",
      "    },\n",
      "    \"10\": {\n",
      "      \"filename\": \"gall6.txt\",\n",
      "      \"md5\": \"e8645e46f3a8ed30fba3ad582cd6bb84\"\n",
      "    },\n",
      "    \"11\": {\n",
      "      \"filename\": \"gall7.txt\",\n",
      "      \"md5\": \"db56d8ca46c77ee856e6c2ea72556579\"\n",
      "    },\n",
      "    \"12\": {\n",
      "      \"filename\": \"gall8.txt\",\n",
      "      \"md5\": \"bfc174bca341688b801e1252b39c4fe8\"\n",
      "    },\n",
      "    \"13\": {\n",
      "      \"filename\": \"eutropius1.txt\",\n",
      "      \"md5\": \"0dba231b290b45abd77a5c49c95f5a27\"\n",
      "    },\n",
      "    \"14\": {\n",
      "      \"filename\": \"eutropius10.txt\",\n",
      "      \"md5\": \"d41751aec253d7168bd0bb7ff961c794\"\n",
      "    },\n",
      "    \"15\": {\n",
      "      \"filename\": \"eutropius2.txt\",\n",
      "      \"md5\": \"40600e2dcf920c695eb4cac484635a7a\"\n",
      "    },\n",
      "    \"16\": {\n",
      "      \"filename\": \"eutropius3.txt\",\n",
      "      \"md5\": \"1b2a672fc1432c5dac2b4283f072b338\"\n",
      "    },\n",
      "    \"17\": {\n",
      "      \"filename\": \"eutropius4.txt\",\n",
      "      \"md5\": \"d76caf78179cc29fdc6847010d0b76e8\"\n",
      "    },\n",
      "    \"18\": {\n",
      "      \"filename\": \"eutropius5.txt\",\n",
      "      \"md5\": \"d63d9bfef33046d4d047319b4b004a84\"\n",
      "    },\n",
      "    \"19\": {\n",
      "      \"filename\": \"eutropius6.txt\",\n",
      "      \"md5\": \"3235ad10251940c0119bc257625274c3\"\n",
      "    },\n",
      "    \"20\": {\n",
      "      \"filename\": \"eutropius7.txt\",\n",
      "      \"md5\": \"731dc1b29e08b7261b0c165dcfcfb793\"\n",
      "    },\n",
      "    \"21\": {\n",
      "      \"filename\": \"eutropius8.txt\",\n",
      "      \"md5\": \"16033e61f59202679367a1bb7d5919e3\"\n",
      "    },\n",
      "    \"22\": {\n",
      "      \"filename\": \"eutropius9.txt\",\n",
      "      \"md5\": \"48bd302b927c6379d697150dd56c9189\"\n",
      "    },\n",
      "    \"23\": {\n",
      "      \"filename\": \"prud1.txt\",\n",
      "      \"md5\": \"57588fae1677a2f206941f38deb345b0\"\n",
      "    },\n",
      "    \"24\": {\n",
      "      \"filename\": \"prud10.txt\",\n",
      "      \"md5\": \"32f15144437f2bd3d9a6ed26ffc76a19\"\n",
      "    },\n",
      "    \"25\": {\n",
      "      \"filename\": \"prud11.txt\",\n",
      "      \"md5\": \"1286ea56d5390e69674e0c2dea8e3ea4\"\n",
      "    },\n",
      "    \"26\": {\n",
      "      \"filename\": \"prud12.txt\",\n",
      "      \"md5\": \"b15f7a1c485e0084e8f3d32c0174766e\"\n",
      "    },\n",
      "    \"27\": {\n",
      "      \"filename\": \"prud13.txt\",\n",
      "      \"md5\": \"b8b4d17ee7bf78150a0dc5e0bfdc183a\"\n",
      "    },\n",
      "    \"28\": {\n",
      "      \"filename\": \"prud14.txt\",\n",
      "      \"md5\": \"228712225e89a3f1b49533d0d32da392\"\n",
      "    },\n",
      "    \"29\": {\n",
      "      \"filename\": \"prud2.txt\",\n",
      "      \"md5\": \"38a6f1cc241547c79aeb7a1c25f8509d\"\n",
      "    },\n",
      "    \"30\": {\n",
      "      \"filename\": \"prud3.txt\",\n",
      "      \"md5\": \"bee85a06ab5df2134b9a2b9205c1d965\"\n",
      "    },\n",
      "    \"31\": {\n",
      "      \"filename\": \"prud4.txt\",\n",
      "      \"md5\": \"08871540d4d70703402b9c7ba036d92d\"\n",
      "    },\n",
      "    \"32\": {\n",
      "      \"filename\": \"prud5.txt\",\n",
      "      \"md5\": \"aebf4e7ad7e51bab93d5657daea4d28c\"\n",
      "    },\n",
      "    \"33\": {\n",
      "      \"filename\": \"prud6.txt\",\n",
      "      \"md5\": \"9a4a6e6e75d74cc91add3e3c5a01e59b\"\n",
      "    },\n",
      "    \"34\": {\n",
      "      \"filename\": \"prud7.txt\",\n",
      "      \"md5\": \"966108346530f9b55bdc42532d9d8595\"\n",
      "    },\n",
      "    \"35\": {\n",
      "      \"filename\": \"prud8.txt\",\n",
      "      \"md5\": \"7a16e3673f6c115a8fdf8d6a0d6a0ea3\"\n",
      "    },\n",
      "    \"36\": {\n",
      "      \"filename\": \"prud9.txt\",\n",
      "      \"md5\": \"3e12bd972124a4507b727a69e1af91cd\"\n",
      "    },\n",
      "    \"37\": {\n",
      "      \"filename\": \"plato__alcibiades-1__grc.json\",\n",
      "      \"md5\": \"751fdf9b9626863d5dacacb6509d4a40\"\n",
      "    },\n",
      "    \"38\": {\n",
      "      \"filename\": \"plato__alcibiades-2__grc.json\",\n",
      "      \"md5\": \"dd4b96417126597098b8a96af2944e13\"\n",
      "    },\n",
      "    \"39\": {\n",
      "      \"filename\": \"plato__apology__grc.json\",\n",
      "      \"md5\": \"e4b546c37bac017b9890d627ec1043ff\"\n",
      "    },\n",
      "    \"40\": {\n",
      "      \"filename\": \"plato__charmides__grc.json\",\n",
      "      \"md5\": \"3c61ef4f4ef3e3af8072e91db49a83d0\"\n",
      "    },\n",
      "    \"41\": {\n",
      "      \"filename\": \"plato__cratylus__grc.json\",\n",
      "      \"md5\": \"51b914a758394750b76732ea39491eb2\"\n",
      "    },\n",
      "    \"42\": {\n",
      "      \"filename\": \"plato__crito__grc.json\",\n",
      "      \"md5\": \"b0cd1626398dd3afbf908bfe17d50ed7\"\n",
      "    },\n",
      "    \"43\": {\n",
      "      \"filename\": \"plato__euthydemus__grc.json\",\n",
      "      \"md5\": \"74adeebf47d56c53c7a82883c66d18c0\"\n",
      "    },\n",
      "    \"44\": {\n",
      "      \"filename\": \"plato__euthyphro__grc.json\",\n",
      "      \"md5\": \"dc6b9eaca2f05d9467e3e655bbd27d38\"\n",
      "    },\n",
      "    \"45\": {\n",
      "      \"filename\": \"plato__hipparchus__grc.json\",\n",
      "      \"md5\": \"e913d77f04e17a4dcd051fcec55bb25a\"\n",
      "    },\n",
      "    \"46\": {\n",
      "      \"filename\": \"plato__laches__grc.json\",\n",
      "      \"md5\": \"9ae676aa76485db29ce62abf2316b44d\"\n",
      "    },\n",
      "    \"47\": {\n",
      "      \"filename\": \"plato__lovers__grc.json\",\n",
      "      \"md5\": \"ef4ba45a873378b971d3abcdcc6d56d7\"\n",
      "    },\n",
      "    \"48\": {\n",
      "      \"filename\": \"plato__lysis__grc.json\",\n",
      "      \"md5\": \"0c0c5e68524d10248a8cf3a9e62a6c4d\"\n",
      "    },\n",
      "    \"49\": {\n",
      "      \"filename\": \"plato__parmenides__grc.json\",\n",
      "      \"md5\": \"418c9f8c21ac0f1af65dedbc71938a0f\"\n",
      "    },\n",
      "    \"50\": {\n",
      "      \"filename\": \"plato__phaedo__grc.json\",\n",
      "      \"md5\": \"f593b5b5c114982fd1fda8779f7158d7\"\n",
      "    },\n",
      "    \"51\": {\n",
      "      \"filename\": \"plato__phaedrus__grc.json\",\n",
      "      \"md5\": \"7ddcaf234a08dfb6a87e0e168540833e\"\n",
      "    },\n",
      "    \"52\": {\n",
      "      \"filename\": \"plato__philebus__grc.json\",\n",
      "      \"md5\": \"f5098a2471ec46c33a4861e7c86d9d39\"\n",
      "    },\n",
      "    \"53\": {\n",
      "      \"filename\": \"plato__sophist__grc.json\",\n",
      "      \"md5\": \"ab4cbdbdcdeca510d230396151302d71\"\n",
      "    },\n",
      "    \"54\": {\n",
      "      \"filename\": \"plato__statesman__grc.json\",\n",
      "      \"md5\": \"9707acc3a72834aa63017a743bb3afec\"\n",
      "    },\n",
      "    \"55\": {\n",
      "      \"filename\": \"plato__symposium__grc.json\",\n",
      "      \"md5\": \"eafbf58e7624372a71e2cdd0d39d2525\"\n",
      "    },\n",
      "    \"56\": {\n",
      "      \"filename\": \"plato__theaetetus__grc.json\",\n",
      "      \"md5\": \"fd50f4a9bbdaf4ab7ccf630c5d9db223\"\n",
      "    },\n",
      "    \"57\": {\n",
      "      \"filename\": \"plato__theages__grc.json\",\n",
      "      \"md5\": \"53d891aa6cc8cce2eb83cc6d3ec35a18\"\n",
      "    },\n",
      "    \"58\": {\n",
      "      \"filename\": \"homer__odyssey__grc.json\",\n",
      "      \"md5\": \"f46270638106d2daaa0a2828cf268422\"\n",
      "    },\n",
      "    \"59\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-1-to-dionysus__grc.json\",\n",
      "      \"md5\": \"ff88ac17a6adbeebe517971e576f6e27\"\n",
      "    },\n",
      "    \"60\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-10-to-aphrodite__grc.json\",\n",
      "      \"md5\": \"9bbe5c96b81dd9660f4e0a8be0662ea6\"\n",
      "    },\n",
      "    \"61\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-11-to-athena__grc.json\",\n",
      "      \"md5\": \"d3bbcacc93dbc839579a8717a46ad3e3\"\n",
      "    },\n",
      "    \"62\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-12-to-hera__grc.json\",\n",
      "      \"md5\": \"d7248209627c5b489b63019665558cfe\"\n",
      "    },\n",
      "    \"63\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-13-to-demeter__grc.json\",\n",
      "      \"md5\": \"fa044a41a8fecc6bba0ce18e9f49c13d\"\n",
      "    },\n",
      "    \"64\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-14-to-the-mother-of-the-gods__grc.json\",\n",
      "      \"md5\": \"d4d7d72adb171149d136740b06a7d286\"\n",
      "    },\n",
      "    \"65\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-15-to-heracles__grc.json\",\n",
      "      \"md5\": \"a9c509199cde7bca8405bc4b9e422bd7\"\n",
      "    },\n",
      "    \"66\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-16-to-asclepius__grc.json\",\n",
      "      \"md5\": \"d2ef4f2efdb107980eeceb2f4943c19b\"\n",
      "    },\n",
      "    \"67\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-17-to-the-dioscuri__grc.json\",\n",
      "      \"md5\": \"243ef6d48ed0ee0c7d224cc78bbf6f7e\"\n",
      "    },\n",
      "    \"68\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-18-to-hermes__grc.json\",\n",
      "      \"md5\": \"9f558df2a69cb840bf6f878be0dc49f8\"\n",
      "    },\n",
      "    \"69\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-19-to-pan__grc.json\",\n",
      "      \"md5\": \"5528b4957c582a1176834090cf800d5e\"\n",
      "    },\n",
      "    \"70\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-2-to-demeter__grc.json\",\n",
      "      \"md5\": \"5c75d7ceb1e478998e9cf09caa8296b5\"\n",
      "    },\n",
      "    \"71\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-20-to-hephaestus__grc.json\",\n",
      "      \"md5\": \"4f59a40d728477a5c68ed2865af81ee6\"\n",
      "    },\n",
      "    \"72\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-21-to-apollo__grc.json\",\n",
      "      \"md5\": \"6fb6d54dc1860c619cd824c4f5d86cf7\"\n",
      "    },\n",
      "    \"73\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-22-to-poseidon__grc.json\",\n",
      "      \"md5\": \"a16791d3134f78e99d7ca7fd6b9c42bb\"\n",
      "    },\n",
      "    \"74\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-23-to-zeus__grc.json\",\n",
      "      \"md5\": \"086a9c28056e32599ae7a4b6761fa4e0\"\n",
      "    },\n",
      "    \"75\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-24-to-hestia__grc.json\",\n",
      "      \"md5\": \"fc88b8a9601833a2c21cdcd7d88801b1\"\n",
      "    },\n",
      "    \"76\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-25-to-the-muses-and-apollo__grc.json\",\n",
      "      \"md5\": \"53e24924a14265a3320e479797432523\"\n",
      "    },\n",
      "    \"77\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-26-to-dionysus__grc.json\",\n",
      "      \"md5\": \"34446d2ae6668ee82629b6c47738e815\"\n",
      "    },\n",
      "    \"78\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-27-to-artemis__grc.json\",\n",
      "      \"md5\": \"9cfaad773cb8232718f78c2d0a07306b\"\n",
      "    },\n",
      "    \"79\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-28-to-athena__grc.json\",\n",
      "      \"md5\": \"229ec0914943fd5c848330e268c22105\"\n",
      "    },\n",
      "    \"80\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-29-to-hestia__grc.json\",\n",
      "      \"md5\": \"c7e40337b22b15f623d0b6d54bc82817\"\n",
      "    },\n",
      "    \"81\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-3-to-delian-and-pythian-apollo__grc.json\",\n",
      "      \"md5\": \"e92882a9d1e686860cb5c45bb4f97d9e\"\n",
      "    },\n",
      "    \"82\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-30-to-earth__grc.json\",\n",
      "      \"md5\": \"f663c38411303148ed4d1349141e13c3\"\n",
      "    },\n",
      "    \"83\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-31-to-helios__grc.json\",\n",
      "      \"md5\": \"31300fe836e4086d906ee7bd2033089f\"\n",
      "    },\n",
      "    \"84\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-32-to-selene__grc.json\",\n",
      "      \"md5\": \"a296d0d31380776958e637f08d3dd810\"\n",
      "    },\n",
      "    \"85\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-33-to-the-dioscuri__grc.json\",\n",
      "      \"md5\": \"a64bc1a297a83f974c0a633fe874f789\"\n",
      "    },\n",
      "    \"86\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-4-to-hermes__grc.json\",\n",
      "      \"md5\": \"8a994c0d80e5234823b1e7c4cf2fca0b\"\n",
      "    },\n",
      "    \"87\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-5-to-aphrodite__grc.json\",\n",
      "      \"md5\": \"0e53b784d0a0acc3b18a5c5ee6441d83\"\n",
      "    },\n",
      "    \"88\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-6-to-aphrodite__grc.json\",\n",
      "      \"md5\": \"8ed4d1deef268e3fdfc3a15c4d529665\"\n",
      "    },\n",
      "    \"89\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-7-to-dionysus__grc.json\",\n",
      "      \"md5\": \"8c88adbb3511b074e95b4f9b5e21f91b\"\n",
      "    },\n",
      "    \"90\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-8-to-ares__grc.json\",\n",
      "      \"md5\": \"904d1b6c2d4f20a314b9a89e5433bdec\"\n",
      "    },\n",
      "    \"91\": {\n",
      "      \"filename\": \"homeric-hymns__hymn-9-to-artemis__grc.json\",\n",
      "      \"md5\": \"95ae7ea977ccd0130ac3950aade30db4\"\n",
      "    }\n",
      "  },\n",
      "  \"model_output_file\": \"is_transliterated_greek.mdl.0.21.3.joblib\",\n",
      "  \"model_output_md5\": \"1bfa763b7d8d2ff75c9a88982681ce15\",\n",
      "  \"labels\": [\n",
      "    0,\n",
      "    1\n",
      "  ],\n",
      "  \"best_score\": 0.9709996945896789,\n",
      "  \"best_params\": {\n",
      "    \"max_features\": 0.4,\n",
      "    \"n_estimators\": 300\n",
      "  },\n",
      "  \"comment\": \"Transliterated Greek Classifier\",\n",
      "  \"code_generated_by\": \"loanwords_problems_solutions.ipynb\",\n",
      "  \"feature_encoding_fun\": \"encode_word\",\n",
      "  \"char_encoding_file\": \"latin.greek.char_idx.pkl\",\n",
      "  \"author\": \"Todd Cook\",\n",
      "  \"sample_pipeline\": \"process_latin_text_pipeline.0.21.3.joblib\"\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "print(json.dumps(params, indent=2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('latin.greek.char_idx.pkl', 'wb') as fout:\n",
    "    pickle.dump(char_idx, fout)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reconstitute the classifier for use at runtime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = load(model_output_file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Some demo examples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0])"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier.predict(\n",
    "    sparse.csr_matrix(np.array([encode_word(word, char_idx, max_len) for word in ['quid', 'est', 'veritas']])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 1, 1])"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier.predict(\n",
    "    sparse.csr_matrix(np.array([encode_word(word, char_idx, max_len) for word in 'ou eis panta ton'.split()])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Smoothing Cluster Holes\n",
    "as you can see from the example above, sometime the classifier will drop out and not classify something correctly in a sequence. Quite often these dropouts are words that could be found in either language. If two good classifications bookend a dropout, then the dropout should probably be filled in.  We've got a function for that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1., 1., 1., 1.])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "patch_cluster_holes(classifier.predict(\n",
    "    sparse.csr_matrix(np.array([encode_word(word, char_idx, max_len) for word in 'ou eis panta ton'.split()]))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Now let's look at the author Pliny the Younger for transliterated Greek words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [50:54<00:00, 293.48s/it]\n"
     ]
    }
   ],
   "source": [
    "greek_in_pliny = set()\n",
    "latin_reader = get_corpus_reader(corpus_name='latin_text_latin_library', language='latin')\n",
    "results = defaultdict(list)\n",
    "\n",
    "selected_files = [file for file in latin_reader.fileids()\n",
    "                  if 'pliny.ep' in file]\n",
    "\n",
    "for file in tqdm(selected_files):\n",
    "    for sent in latin_reader.sents(file):\n",
    "        unseen_X = process_latin_text_pipeline.fit_transform([(sent)])\n",
    "        if unseen_X and len(unseen_X[0][0]) > 1:\n",
    "            arr = classifier.predict(\n",
    "                sparse.csr_matrix(np.array([encode_word(word, char_idx, max_len) for sentence in unseen_X for word in\n",
    "                          sentence])))\n",
    "            arr = patch_cluster_holes(arr)\n",
    "            found_greek = extract_words(unseen_X[0], *run_length_encoding(arr))  # works with sent\n",
    "            if found_greek:\n",
    "                results[file].append(found_greek)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[[['gar', 'hoi', 'euzêloi']],\n",
       "  [['kata', 'Meidiou']],\n",
       "  [['akousma', 'epainos']]],\n",
       " [[['logismos', 'de', 'okno', 'ne', 'pherei']],\n",
       "  [['phônê', 'ne', 'kai', 'gegêthôs', 'kai']],\n",
       "  [['keitai', 'Patroklos']],\n",
       "  [['tauta', 'hyper', 'melêsei']],\n",
       "  [['hoti', 'toioutos', 'esti', 'ne', 'hoisper']]],\n",
       " [[['kakos', 'êde', 'kai', 'esthlos']]],\n",
       " [[['nomou', 'Memphitou']]],\n",
       " [[['edôke', 'patêr', 'hetero', 'ne', 'aneneusen']],\n",
       "  [['kai', 'kyaneêsi', 'ne', 'ep', 'ophrysi', 'neuse']],\n",
       "  [['kai', 'gar', 'onar', 'ek', 'Dios', 'esti']],\n",
       "  [['logisamenos',\n",
       "    'illud',\n",
       "    'heis',\n",
       "    'oiônos',\n",
       "    'aristos',\n",
       "    'amynesthai',\n",
       "    'peri']],\n",
       "  [['panta', 'denique', 'litho', 'ne', 'kinô']],\n",
       "  [['pros',\n",
       "    'de',\n",
       "    'autou',\n",
       "    'taxei',\n",
       "    'peithô',\n",
       "    'tis',\n",
       "    'epekathêto',\n",
       "    'toisi',\n",
       "    'cheilesi']],\n",
       "  [['ekêlei', 'kai', 'monos'], ['enkateleipe', 'tois', 'akroômenois']],\n",
       "  [['kai', 'epea', 'niphadessi', 'ne', 'eoikota']]]]"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random.sample(list(results.values()), 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[[['autou', 'tou', 'thêriou', 'êkousate']],\n",
       "  [['Sophokleis', 'uocantur', 'apo', 'tou', 'sophôs', 'kai', 'kaleisthai']],\n",
       "  [['ti', 'diateinomai']]],\n",
       " [[['kakos', 'êde', 'kai', 'esthlos']]],\n",
       " [[['hosiê', 'phthimenoisi']],\n",
       "  [['neoi', 'teirousi', 'machêtai']],\n",
       "  [['amphi', 'de', 'salpinxe', 'ne', 'megas', 'ouranos'],\n",
       "   ['oute', 'thalassês', 'kyma', 'toso']],\n",
       "  [['anthrôpoi', 'miaroi', 'kolakes', 'kai'],\n",
       "   ['ou', 'lithois'],\n",
       "   ['oude', 'plinthois', 'egô'],\n",
       "   ['ouk', 'ek'],\n",
       "   ['Euboia', 'ne', 'probebalesthai'],\n",
       "   ['Ìegô', 'de', 'oimai'],\n",
       "   ['andres',\n",
       "    'Athênaioi',\n",
       "    'tous',\n",
       "    'theous',\n",
       "    'ekeino',\n",
       "    'ne',\n",
       "    'methyei',\n",
       "    'ne',\n",
       "    'megethei']],\n",
       "  [['tote', 'egô'], ['kai', 'pollô', 'reonti', 'kath']],\n",
       "  [['ek', 'pleonexias', 'kai', 'ponêrias', 'tis', 'houtos'],\n",
       "   ['kai', 'mikro'],\n",
       "   ['hapanta', 'anechaitise', 'kai', 'dielyse']],\n",
       "  [['apeschoinismenos', 'hapasi', 'tois', 'en', 'polei', 'dikaiois'],\n",
       "   ['tauta', 'eleo', 'ne', 'proudôkas', 'Aristogeito']],\n",
       "  [['pros', 'hous', 'autos', 'echôsas', 'limenas', 'kai'],\n",
       "   ['pros', 'toutous', 'hormizou']],\n",
       "  [['oudena', 'horô'], ['onta', 'alla', 'panta']],\n",
       "  [['dedoika',\n",
       "    'doxête',\n",
       "    'tisi',\n",
       "    'to',\n",
       "    'ne',\n",
       "    'aei',\n",
       "    'boulomeno',\n",
       "    'ne',\n",
       "    'einai'],\n",
       "   ['polei', 'paidotribei']],\n",
       "  [['oude',\n",
       "    'gar',\n",
       "    'tous',\n",
       "    'progonous',\n",
       "    'hypolambanô',\n",
       "    'ta',\n",
       "    'dikastêria',\n",
       "    'tauta'],\n",
       "   ['oikodomêsai', 'hina', 'tous', 'toioutous', 'en', 'autois']],\n",
       "  [['esti', 'ponêrias', 'kai', 'palinkapêlos', 'kai']],\n",
       "  [['gar',\n",
       "    'andres',\n",
       "    'Athênaioi',\n",
       "    'to',\n",
       "    'auto',\n",
       "    'phthengesthai',\n",
       "    'to',\n",
       "    'ne',\n",
       "    'kai',\n",
       "    'to',\n",
       "    'ne',\n",
       "    'nomonAE',\n",
       "    'hota'],\n",
       "   ['phônê', 'ne', 'aphiê', 'ho', 'nomos', 'hetera']],\n",
       "  [['epeita', 'anaphainetai', 'peri']],\n",
       "  [['all',\n",
       "    'enkathêmenoi',\n",
       "    'kai',\n",
       "    'enedreuontes',\n",
       "    'en',\n",
       "    'akroasei',\n",
       "    'eiselaunete',\n",
       "    'auto'],\n",
       "   ['tous', 'paranomous', 'logous']],\n",
       "  [['all',\n",
       "    'en',\n",
       "    'tais',\n",
       "    'hippodromiais',\n",
       "    'eis',\n",
       "    'to',\n",
       "    'ne',\n",
       "    'tou',\n",
       "    'pragmatos',\n",
       "    'auto',\n",
       "    'ne',\n",
       "    'dromo',\n",
       "    'ne',\n",
       "    'eiselaunete']],\n",
       "  [['helkopoieis', 'kai'], ['soi', 'melei'], ['logô', 'ne', 'poleôs']],\n",
       "  [['ouk', 'apopempsesthe', 'to'],\n",
       "   ['symphora', 'ne', 'syllabontes'],\n",
       "   ['ep', 'onomatôn', 'dia', 'politeias', 'pleonta', 'timôrêsesthe']]],\n",
       " [[['gar', 'aoidê'],\n",
       "   ['epikleious',\n",
       "    'anthrôpoi',\n",
       "    'tis',\n",
       "    'akouontessi',\n",
       "    'neôtatê',\n",
       "    'amphipelêtai']]],\n",
       " [[['ou', 'tis', 'emeu']],\n",
       "  [['kata', 'kephalaio']],\n",
       "  [['hyper', 'Ktêsiphôntos']]]]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random.sample(list(results.values()), 5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Assessing the model \n",
    "We can assess the classifier's performance by showing how it behaves with different amounts of training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "title = \"Learning Curves (RandomForest)\"\n",
    "# Cross validation with 100 iterations to get smoother mean test and train\n",
    "# score curves, each time with 20% data randomly selected as a validation set.\n",
    "cv = ShuffleSplit(n_splits=50, test_size=0.2, random_state=0)  # changed 100 to 50\n",
    "estimator = RandomForestClassifier(**mdl_params)\n",
    "ylim = (0.7, 1.01)\n",
    "n_jobs = 7\n",
    "train_sizes = np.linspace(.1, 1.0, 5)\n",
    "fig = plt.figure(figsize=(7, 7))\n",
    "plt.title(title)\n",
    "if ylim is not None:\n",
    "    plt.ylim(*ylim)\n",
    "plt.xlabel(\"Training examples\")\n",
    "plt.ylabel(\"Score\")\n",
    "train_sizes, train_scores, test_scores = learning_curve(\n",
    "    estimator, all_X, all_y, cv=cv, n_jobs=n_jobs, train_sizes=train_sizes)\n",
    "train_scores_mean = np.mean(train_scores, axis=1)\n",
    "train_scores_std = np.std(train_scores, axis=1)\n",
    "test_scores_mean = np.mean(test_scores, axis=1)\n",
    "test_scores_std = np.std(test_scores, axis=1)\n",
    "plt.grid()\n",
    "plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
    "                 train_scores_mean + train_scores_std, alpha=0.1,\n",
    "                 color=\"r\")\n",
    "plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
    "                 test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
    "plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
    "         label=\"Training score\")\n",
    "plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
    "         label=\"Cross-validation score\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.tight_layout(pad=0.5, w_pad=20, h_pad=0.5)\n",
    "plt.show()\n",
    "# plt.savefig('loanwords.prob.solutions.learningcurve.png')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Appendix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Find which texts in the Latin Library have the most transliterated Greek"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 3/2141 [00:21<4:11:16,  7.05s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 30 for perseuerantiamtantasustinentem, truncating to 25\n",
      "  1%|▏         | 27/2141 [01:20<1:04:19,  1.83s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for repromissionesobturauerunt, truncating to 25\n",
      "  1%|▏         | 28/2141 [01:22<1:04:29,  1.83s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for impudentioremintemperantia, truncating to 25\n",
      "  1%|▏         | 30/2141 [01:28<1:31:01,  2.59s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for iniuriarumquetollerabilium, truncating to 25\n",
      " 32%|███▏      | 684/2141 [24:11<51:33,  2.12s/files]  WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 27 for souramagnificentissimamente, truncating to 25\n",
      " 36%|███▌      | 766/2141 [26:55<32:02,  1.40s/files]  WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for appropinquantibusquadrigis, truncating to 25\n",
      " 36%|███▌      | 768/2141 [27:00<40:38,  1.78s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for recollegissetanimaduertisset, truncating to 25\n",
      " 38%|███▊      | 818/2141 [28:43<32:26,  1.47s/files]  WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 73 for fa3RuLCBxdWlhIGRlIGl1cmUgbm9uIHBudGUyYXQsIHZlbmllbnMsIGUiY2xlc2lhbSBpcHNh, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 71 for GksIGZyYXRlcm5pdGF0ZW0gdHUhbSBwbGUuaXUzIGluc3RydWUtdXMuIFZlcnUtIHRhbWUu, truncating to 25\n",
      " 66%|██████▌   | 1412/2141 [51:33<38:37,  3.18s/files]  WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for Anebanensisepiscopidebitae, truncating to 25\n",
      " 72%|███████▏  | 1540/2141 [54:59<22:52,  2.28s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for Thensaurochrysonicochrysides, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for Thensaurochrysonicochrysides, truncating to 25\n",
      " 72%|███████▏  | 1550/2141 [55:25<27:03,  2.75s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 48 for Ythalonimualoniuthsicorathiisthymhimihymacomsyth, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for combaepumamitalmetlotiambeat, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for iulecantheconaalonimbalumbar, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for hunesobinesubicsillimbalim, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 32 for esseantidamossonalemuedubertefet, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 33 for donobunhuneccilthumucommucroluful, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 36 for altanimauosduberithemhuarcharistolem, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 28 for sittesedanecnasotersahelicot, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 29 for alemusdubertimurmucopsuistiti, truncating to 25\n",
      "WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 39 for aoccaaneclictorbodesiussilimlimmimcolus, truncating to 25\n",
      " 77%|███████▋  | 1650/2141 [58:43<22:59,  2.81s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for MahummicolarumChristicolas, truncating to 25\n",
      " 81%|████████  | 1732/2141 [1:01:21<16:17,  2.39s/files]WARNING:mlyoucanuse.featurize_text_fun:Excessive word length 26 for conprehendimustemperantiam, truncating to 25\n",
      "100%|██████████| 2141/2141 [1:15:06<00:00,  2.69s/files]\n"
     ]
    }
   ],
   "source": [
    "corpus_files = {}\n",
    "latin_reader = get_corpus_reader(corpus_name='latin_text_latin_library', language='latin')\n",
    "greek_in_corpus_selection = set()\n",
    "found_greek = []\n",
    "\n",
    "for full_path in tqdm(latin_reader.fileids(), total=len(latin_reader.fileids()), unit='files'):\n",
    "    filename = full_path[full_path.rfind('/') + 1:]\n",
    "    unseen_X = process_latin_text_pipeline.fit_transform(list(latin_reader.sents([full_path])))\n",
    "    distinct_unseen = distinct_words(unseen_X)\n",
    "    unseen_words = list(distinct_unseen)\n",
    "    total_words = [word\n",
    "                   for sentence in unseen_X\n",
    "                   for word in sentence]\n",
    "    arr = classifier.predict(sparse.csr_matrix(np.array([encode_word(word, char_idx, max_len) for word in total_words])))\n",
    "    total_greek_words = np.count_nonzero(arr)\n",
    "    marks = arr.tolist()\n",
    "    if marks:\n",
    "        found_greek = [total_words[idx]\n",
    "                       for idx, point in enumerate(marks)\n",
    "                       if point == 1]\n",
    "        greek_in_corpus_selection |= set(found_greek)\n",
    "        corpus_files[filename] = (len(total_words), total_greek_words)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Greek words found in Latin selection 16,612\n",
      "Number of Greek words not in training data: 14,304\n",
      "Random sample: ['gelôtopoio', 'Andr', 'constantinopolitana', 'solummagna', 'anhelas', 'Lezenais', 'Colophone', 'prias', 'Strete', 'metrocomias', 'stephephoros', 'peritretos', 'intelligentiores', 'seriò', 'etumos', 'Athamantidos', 'apodixin', 'pie', 'purous', 'mag']\n"
     ]
    }
   ],
   "source": [
    "print(f'Total Greek words found in Latin selection {len(greek_in_corpus_selection):,}')\n",
    "print(f'Number of Greek words not in training data: {len(greek_in_corpus_selection - only_greek_transliterated):,}')\n",
    "print(f'Random sample: {random.sample(greek_in_corpus_selection, 20)}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2039\n",
      "('preface.txt', 411, 0, 0.0)\n",
      "('serm18.txt', 1363, 0, 0.0)\n",
      "('balde1.txt', 109, 0, 0.0)\n",
      "('gall5.txt', 7489, 0, 0.0)\n",
      "('legagr3.txt', 1249, 0, 0.0)\n",
      "('repub4.txt', 202, 0, 0.0)\n",
      "('dicquid.txt', 52, 0, 0.0)\n",
      "('dulcesolum.txt', 73, 0, 0.0)\n",
      "('dumdomus.txt', 46, 0, 0.0)\n",
      "('dumestas.txt', 13, 0, 0.0)\n",
      "('eutropius1.txt', 1569, 0, 0.0)\n",
      "('eutropius9.txt', 2083, 0, 0.0)\n",
      "('fortunat.txt', 101, 0, 0.0)\n",
      "('contro.txt', 926, 0, 0.0)\n",
      "('qualitate.txt', 353, 0, 0.0)\n",
      "('gaud.txt', 113, 0, 0.0)\n",
      "('janus2.txt', 2, 0, 0.0)\n",
      "('jfkhonor.txt', 6, 0, 0.0)\n",
      "('liv.per.txt', 2, 0, 0.0)\n",
      "('liv.per1.txt', 444, 0, 0.0)\n",
      "('liv.per10.txt', 117, 0, 0.0)\n",
      "('liv.per101.txt', 77, 0, 0.0)\n",
      "('liv.per102.txt', 97, 0, 0.0)\n",
      "('liv.per106.txt', 111, 0, 0.0)\n",
      "('liv.per108.txt', 81, 0, 0.0)\n",
      "('liv.per11.txt', 198, 0, 0.0)\n",
      "('liv.per110.txt', 121, 0, 0.0)\n",
      "('liv.per111.txt', 106, 0, 0.0)\n",
      "('liv.per112.txt', 129, 0, 0.0)\n",
      "('liv.per114.txt', 92, 0, 0.0)\n",
      "('liv.per115.txt', 77, 0, 0.0)\n",
      "('liv.per117.txt', 125, 0, 0.0)\n",
      "('liv.per118.txt', 72, 0, 0.0)\n",
      "('liv.per119.txt', 151, 0, 0.0)\n",
      "('liv.per12.txt', 93, 0, 0.0)\n",
      "('liv.per120.txt', 135, 0, 0.0)\n",
      "('liv.per121.txt', 42, 0, 0.0)\n",
      "('liv.per122.txt', 35, 0, 0.0)\n",
      "('liv.per123.txt', 66, 0, 0.0)\n",
      "('liv.per124.txt', 73, 0, 0.0)\n",
      "('liv.per125.txt', 69, 0, 0.0)\n",
      "('liv.per126.txt', 44, 0, 0.0)\n",
      "('liv.per127.txt', 105, 0, 0.0)\n",
      "('liv.per128.txt', 50, 0, 0.0)\n",
      "('liv.per129.txt', 80, 0, 0.0)\n",
      "('liv.per13.txt', 166, 0, 0.0)\n",
      "('liv.per130.txt', 74, 0, 0.0)\n",
      "('liv.per131.txt', 64, 0, 0.0)\n",
      "('liv.per132.txt', 67, 0, 0.0)\n",
      "('liv.per133.txt', 78, 0, 0.0)\n",
      "('liv.per134.txt', 46, 0, 0.0)\n",
      "('liv.per135.txt', 16, 0, 0.0)\n",
      "('liv.per136-7.txt', 4, 0, 0.0)\n",
      "('liv.per138.txt', 18, 0, 0.0)\n",
      "('liv.per14.txt', 117, 0, 0.0)\n",
      "('liv.per140.txt', 34, 0, 0.0)\n",
      "('liv.per141.txt', 45, 0, 0.0)\n",
      "('liv.per142.txt', 50, 0, 0.0)\n",
      "('liv.per15.txt', 65, 0, 0.0)\n",
      "('liv.per16.txt', 82, 0, 0.0)\n",
      "('liv.per17.txt', 115, 0, 0.0)\n",
      "('liv.per18.txt', 160, 0, 0.0)\n",
      "('liv.per19.txt', 193, 0, 0.0)\n",
      "('liv.per2.txt', 453, 0, 0.0)\n",
      "('liv.per20.txt', 164, 0, 0.0)\n",
      "('liv.per21.txt', 127, 0, 0.0)\n",
      "('liv.per22.txt', 364, 0, 0.0)\n",
      "('liv.per23.txt', 196, 0, 0.0)\n",
      "('liv.per24.txt', 141, 0, 0.0)\n",
      "('liv.per25.txt', 215, 0, 0.0)\n",
      "('liv.per26.txt', 172, 0, 0.0)\n",
      "('liv.per27.txt', 176, 0, 0.0)\n",
      "('liv.per28.txt', 272, 0, 0.0)\n",
      "('liv.per29.txt', 378, 0, 0.0)\n",
      "('liv.per32.txt', 108, 0, 0.0)\n",
      "('liv.per33.txt', 107, 0, 0.0)\n",
      "('liv.per36.txt', 55, 0, 0.0)\n",
      "('liv.per37.txt', 142, 0, 0.0)\n",
      "('liv.per4.txt', 187, 0, 0.0)\n",
      "('liv.per40.txt', 200, 0, 0.0)\n",
      "('liv.per43.txt', 79, 0, 0.0)\n",
      "('liv.per44.txt', 141, 0, 0.0)\n",
      "('liv.per45.txt', 187, 0, 0.0)\n",
      "('liv.per50.txt', 306, 0, 0.0)\n",
      "('liv.per51.txt', 158, 0, 0.0)\n",
      "('liv.per52.txt', 272, 0, 0.0)\n",
      "('liv.per53.txt', 41, 0, 0.0)\n",
      "('liv.per54.txt', 128, 0, 0.0)\n",
      "('liv.per57.txt', 173, 0, 0.0)\n",
      "('liv.per58.txt', 198, 0, 0.0)\n",
      "('liv.per60.txt', 206, 0, 0.0)\n",
      "('liv.per62.txt', 57, 0, 0.0)\n",
      "('liv.per63.txt', 69, 0, 0.0)\n",
      "('liv.per64.txt', 111, 0, 0.0)\n",
      "('liv.per65.txt', 72, 0, 0.0)\n",
      "('liv.per66.txt', 41, 0, 0.0)\n",
      "('liv.per68.txt', 164, 0, 0.0)\n",
      "('liv.per69.txt', 137, 0, 0.0)\n",
      "('liv.per70.txt', 151, 0, 0.0)\n",
      "('liv.per71.txt', 83, 0, 0.0)\n",
      "('liv.per72.txt', 70, 0, 0.0)\n",
      "('liv.per73.txt', 127, 0, 0.0)\n",
      "('liv.per74.txt', 88, 0, 0.0)\n",
      "('liv.per75.txt', 96, 0, 0.0)\n",
      "('liv.per76.txt', 87, 0, 0.0)\n",
      "('liv.per79.txt', 131, 0, 0.0)\n",
      "('liv.per8.txt', 232, 0, 0.0)\n",
      "('liv.per82.txt', 76, 0, 0.0)\n",
      "('liv.per83.txt', 139, 0, 0.0)\n",
      "('liv.per84.txt', 115, 0, 0.0)\n",
      "('liv.per85.txt', 98, 0, 0.0)\n",
      "('liv.per87.txt', 34, 0, 0.0)\n",
      "('liv.per88.txt', 125, 0, 0.0)\n",
      "('liv.per89.txt', 247, 0, 0.0)\n",
      "('liv.per90.txt', 69, 0, 0.0)\n",
      "('liv.per91.txt', 40, 0, 0.0)\n",
      "('liv.per92.txt', 75, 0, 0.0)\n",
      "('liv.per94.txt', 43, 0, 0.0)\n",
      "('liv.per95.txt', 59, 0, 0.0)\n",
      "('liv.per96.txt', 100, 0, 0.0)\n",
      "('liv.per97.txt', 98, 0, 0.0)\n",
      "('liv.per98.txt', 116, 0, 0.0)\n",
      "('liv.pr.txt', 483, 0, 0.0)\n",
      "('maytitle.txt', 43, 0, 0.0)\n",
      "('mir6.txt', 49, 0, 0.0)\n",
      "('prec.terr.txt', 220, 0, 0.0)\n",
      "('precatio.txt', 123, 0, 0.0)\n",
      "('prud12.txt', 404, 0, 0.0)\n",
      "('prud7.txt', 313, 0, 0.0)\n",
      "('prud9.txt', 629, 0, 0.0)\n",
      "('pulchracomis.txt', 33, 0, 0.0)\n",
      "('sulpicia.txt', 262, 0, 0.0)\n",
      "('tevigilans.txt', 51, 0, 0.0)\n",
      "('valesianus.txt', 11, 0, 0.0)\n",
      "('ver1.txt', 4678, 1, 0.00021376656690893543)\n",
      "('gall1.txt', 8158, 2, 0.00024515812699190976)\n",
      "('gall3.txt', 3637, 1, 0.00027495188342040145)\n",
      "('gall8.txt', 6550, 2, 0.0003053435114503817)\n",
      "('cat3.txt', 3212, 1, 0.00031133250311332503)\n",
      "('arch.txt', 3100, 1, 0.0003225806451612903)\n",
      "('serm99.txt', 2699, 1, 0.0003705075954057058)\n",
      "('postreditum2.txt', 2642, 1, 0.0003785011355034065)\n",
      "('gall4.txt', 4563, 2, 0.00043830813061582295)\n",
      "('eutropius7.txt', 2236, 1, 0.0004472271914132379)\n",
      "('bc3.txt', 15164, 7, 0.0004616196254286468)\n",
      "('eutropius6.txt', 2079, 1, 0.000481000481000481)\n",
      "('varia10.txt', 8260, 4, 0.00048426150121065375)\n",
      "('deio.txt', 3916, 2, 0.0005107252298263534)\n",
      "('gall7.txt', 11608, 6, 0.0005168849069607167)\n",
      "('capellanus2.txt', 9497, 5, 0.0005264820469621986)\n",
      "('eutropius8.txt', 1872, 1, 0.0005341880341880342)\n",
      "('eutropius4.txt', 1800, 1, 0.0005555555555555556)\n",
      "('eutropius10.txt', 1798, 1, 0.0005561735261401557)\n",
      "('prov.txt', 5068, 3, 0.0005919494869771113)\n",
      "('varia6.txt', 8297, 5, 0.000602627455706882)\n",
      "('aqua2.txt', 6459, 4, 0.0006192909119058678)\n",
      "('didiul.txt', 1595, 1, 0.0006269592476489029)\n",
      "('cluentio.txt', 20523, 13, 0.0006334356575549384)\n",
      "('haruspicum.txt', 7530, 5, 0.0006640106241699867)\n",
      "('cat4.txt', 2910, 2, 0.0006872852233676976)\n",
      "('liv.8.txt', 13066, 9, 0.0006888106536047758)\n",
      "('quintilian.decl.mai2.txt', 4327, 3, 0.0006933210076265311)\n",
      "('serm5.txt', 2844, 2, 0.0007032348804500703)\n",
      "('gall2.txt', 4195, 3, 0.0007151370679380214)\n",
      "('gall6.txt', 5586, 4, 0.0007160759040458289)\n",
      "('quintilian.decl.mai7.txt', 2750, 2, 0.0007272727272727272)\n",
      "('bc1.txt', 10990, 8, 0.0007279344858962694)\n",
      "('declaratio.txt', 1356, 1, 0.0007374631268436578)\n",
      "('phil10.txt', 2705, 2, 0.0007393715341959334)\n",
      "('liv.6.txt', 13487, 10, 0.0007414547341884778)\n",
      "('amic.txt', 9353, 7, 0.0007484229658932962)\n",
      "('serm95.txt', 1335, 1, 0.000749063670411985)\n",
      "('sall.1.txt', 10662, 8, 0.0007503282686175201)\n",
      "('quintilian.decl.mai15.txt', 2618, 2, 0.0007639419404125286)\n",
      "('42.txt', 1302, 1, 0.0007680491551459293)\n",
      "('tac.ann16.txt', 3897, 3, 0.0007698229407236335)\n",
      "('phil5.txt', 5130, 4, 0.0007797270955165692)\n",
      "('q1.53.txt', 2553, 2, 0.0007833920877399138)\n",
      "('partitione.txt', 9893, 8, 0.0008086525826341858)\n",
      "('hisp.txt', 6038, 5, 0.0008280887711162637)\n",
      "('sall.2.txt', 21442, 18, 0.0008394739296707397)\n",
      "('alex.txt', 10575, 9, 0.000851063829787234)\n",
      "('seneca.ep8.txt', 7019, 6, 0.0008548226243054566)\n",
      "('liv.25.txt', 14686, 13, 0.000885196786054746)\n",
      "('liv.28.txt', 16850, 15, 0.0008902077151335311)\n",
      "('gaius1.txt', 8804, 8, 0.0009086778736937755)\n",
      "('gaius2.txt', 12078, 11, 0.0009107468123861566)\n",
      "('postreditum.txt', 4345, 4, 0.0009205983889528193)\n",
      "('liv.9.txt', 16165, 15, 0.0009279307145066501)\n",
      "('quintilian.decl.mai16.txt', 2133, 2, 0.0009376465072667605)\n",
      "('apuleius7.txt', 4202, 4, 0.0009519276534983341)\n",
      "('varia8.txt', 10438, 10, 0.0009580379383023567)\n",
      "('fam4.txt', 6217, 6, 0.0009650957053241113)\n",
      "('sen.clem.txt', 8245, 8, 0.0009702850212249849)\n",
      "('orosius5.txt', 11277, 11, 0.0009754367296266737)\n",
      "('sen.prov.txt', 4096, 4, 0.0009765625)\n",
      "('eutropius5.txt', 1020, 1, 0.000980392156862745)\n",
      "('varia9.txt', 9175, 9, 0.0009809264305177111)\n",
      "('serm16.txt', 1010, 1, 0.0009900990099009901)\n",
      "('phaedr2.txt', 998, 1, 0.001002004008016032)\n",
      "('gaius4.txt', 9949, 10, 0.001005126143330988)\n",
      "('tac.hist1.txt', 11898, 12, 0.0010085728693898135)\n",
      "('varia.praef.txt', 983, 1, 0.001017293997965412)\n",
      "('histrom7.txt', 2916, 3, 0.00102880658436214)\n",
      "('trin11.txt', 5828, 6, 0.001029512697323267)\n",
      "('trin6.txt', 3859, 4, 0.0010365379632029023)\n",
      "('calpurniusflaccus.txt', 6751, 7, 0.0010368834246778254)\n",
      "('caecilium.txt', 5762, 6, 0.0010413051023950017)\n",
      "('liv.10.txt', 15286, 16, 0.0010467094073007982)\n",
      "('liv.41.txt', 7612, 8, 0.0010509721492380452)\n",
      "('histrom10.txt', 1899, 2, 0.00105318588730911)\n",
      "('q1.80.txt', 943, 1, 0.0010604453870625664)\n",
      "('quintilian.decl.mai9.txt', 4703, 5, 0.00106315118009781)\n",
      "('histrom8.txt', 1877, 2, 0.0010655301012253596)\n",
      "('tac.ann6.txt', 7504, 8, 0.0010660980810234541)\n",
      "('liv.4.txt', 16830, 18, 0.0010695187165775401)\n",
      "('tac.ann4.txt', 10157, 11, 0.0010829969479176922)\n",
      "('phil1.txt', 3688, 4, 0.0010845986984815619)\n",
      "('pontano.txt', 910, 1, 0.001098901098901099)\n",
      "('verres.2.1.txt', 15461, 17, 0.001099540780027165)\n",
      "('epist.txt', 904, 1, 0.0011061946902654867)\n",
      "('varia7.txt', 8056, 9, 0.0011171797418073487)\n",
      "('trin10.txt', 5327, 6, 0.0011263375258119017)\n",
      "('31.txt', 1761, 2, 0.001135718341851221)\n",
      "('sestio.txt', 16722, 19, 0.0011362277239564645)\n",
      "('repub3.txt', 2605, 3, 0.0011516314779270633)\n",
      "('liv.2.txt', 18009, 21, 0.0011660836248542396)\n",
      "('serm73.txt', 855, 1, 0.0011695906432748538)\n",
      "('liv.26.txt', 17088, 20, 0.0011704119850187266)\n",
      "('nahum.txt', 853, 1, 0.0011723329425556857)\n",
      "('liv.3.txt', 20397, 24, 0.0011766436240623621)\n",
      "('histrom3.txt', 2549, 3, 0.0011769321302471558)\n",
      "('eutropius3.txt', 1690, 2, 0.001183431952662722)\n",
      "('fam6.txt', 7599, 9, 0.0011843663639952626)\n",
      "('cat1.txt', 3355, 4, 0.0011922503725782414)\n",
      "('victor.ill.txt', 9205, 11, 0.0011950027159152634)\n",
      "('imp.txt', 6694, 8, 0.0011951000896325067)\n",
      "('ben3.txt', 6687, 8, 0.001196351129056378)\n",
      "('sall.macer.txt', 835, 1, 0.0011976047904191617)\n",
      "('liv.22.txt', 17401, 21, 0.0012068271938394345)\n",
      "('sall.lep.txt', 824, 1, 0.0012135922330097086)\n",
      "('phil9.txt', 1642, 2, 0.001218026796589525)\n",
      "('fragmentumlaurentianum.txt', 817, 1, 0.0012239902080783353)\n",
      "('tac.hist4.txt', 12997, 16, 0.0012310533199969224)\n",
      "('sen.consolatione1.txt', 5681, 7, 0.0012321774335504313)\n",
      "('pliny.panegyricus.txt', 19467, 24, 0.0012328556017876407)\n",
      "('tac.ann11.txt', 4755, 6, 0.0012618296529968455)\n",
      "('histrom9.txt', 2342, 3, 0.0012809564474807857)\n",
      "('tac.ann3.txt', 9352, 12, 0.0012831479897348161)\n",
      "('cat2.txt', 3107, 4, 0.001287415513356936)\n",
      "('liv.29.txt', 12385, 16, 0.0012918853451756157)\n",
      "('quinc.txt', 8498, 11, 0.0012944222169922334)\n",
      "('nepos.eum.txt', 2314, 3, 0.0012964563526361278)\n",
      "('sulla.txt', 9234, 12, 0.001299545159194282)\n",
      "('vegetius1.txt', 4605, 6, 0.0013029315960912053)\n",
      "('ira.txt', 2299, 3, 0.0013049151805132667)\n",
      "('tac.agri.txt', 6851, 9, 0.0013136768354984675)\n",
      "('liv.1.txt', 17350, 23, 0.0013256484149855908)\n",
      "('lamentations.txt', 2259, 3, 0.0013280212483399733)\n",
      "('pesc.txt', 2253, 3, 0.0013315579227696406)\n",
      "('murena.txt', 10493, 14, 0.00133422281521014)\n",
      "('institutes3.txt', 14126, 19, 0.0013450375194676482)\n",
      "('liv.5.txt', 16315, 22, 0.0013484523444682807)\n",
      "('varia11.txt', 9636, 13, 0.0013491075134910752)\n",
      "('institutes4.txt', 13213, 18, 0.0013622947097555438)\n",
      "('tac.ann1.txt', 10898, 15, 0.0013763993393283171)\n",
      "('trin8.txt', 5802, 8, 0.0013788348845225785)\n",
      "('domo.txt', 15140, 21, 0.0013870541611624835)\n",
      "('nepos.att.txt', 3594, 5, 0.0013912075681691708)\n",
      "('verres.2.4.txt', 17197, 24, 0.0013955922544629878)\n",
      "('suet.cal.txt', 7856, 11, 0.00140020366598778)\n",
      "('histrom5.txt', 1427, 2, 0.001401541695865452)\n",
      "('quintilian.decl.mai4.txt', 4274, 6, 0.0014038371548900327)\n",
      "('adbrutum1.txt', 7809, 11, 0.0014086310667178897)\n",
      "('quintilian.institutio11.txt', 15555, 22, 0.0014143362262937962)\n",
      "('verres.2.2.txt', 18934, 27, 0.00142600612654484)\n",
      "('nithardus1.txt', 2789, 4, 0.0014342058085335247)\n",
      "('sen.ira2.txt', 7660, 11, 0.0014360313315926892)\n",
      "('trin7.txt', 6264, 9, 0.0014367816091954023)\n",
      "('caecina.txt', 10421, 15, 0.0014394012090970156)\n",
      "('nepos.han.txt', 2084, 3, 0.0014395393474088292)\n",
      "('phil4.txt', 1365, 2, 0.0014652014652014652)\n",
      "('psplato.virtu.txt', 1364, 2, 0.001466275659824047)\n",
      "('legagr2.txt', 10188, 15, 0.0014723203769140165)\n",
      "('phil14.txt', 3395, 5, 0.0014727540500736377)\n",
      "('trin15.txt', 17651, 26, 0.001473004362359073)\n",
      "('prud5.txt', 2031, 3, 0.0014771048744460858)\n",
      "('tac.hist3.txt', 11402, 17, 0.0014909664971057709)\n",
      "('quintilian.decl.mai12.txt', 6029, 9, 0.0014927848731132857)\n",
      "('lhomond.viris.txt', 28100, 42, 0.001494661921708185)\n",
      "('liv.27.txt', 17376, 26, 0.001496316758747698)\n",
      "('septem.txt', 1335, 2, 0.00149812734082397)\n",
      "('quintilian.decl.mai19.txt', 3335, 5, 0.0014992503748125937)\n",
      "('serm12.txt', 2667, 4, 0.0014998125234345708)\n",
      "('liv.7.txt', 13288, 20, 0.0015051173991571343)\n",
      "('bellafr.txt', 13194, 20, 0.0015158405335758678)\n",
      "('lucretius2.txt', 7883, 12, 0.001522263097805404)\n",
      "('liv.23.txt', 14919, 23, 0.001541658288089014)\n",
      "('histrom14.txt', 1937, 3, 0.0015487867836861124)\n",
      "('trin13.txt', 9675, 15, 0.0015503875968992248)\n",
      "('vatin.txt', 4481, 7, 0.0015621513055121626)\n",
      "('10.txt', 10855, 17, 0.0015660985720865961)\n",
      "('seb.txt', 1277, 2, 0.0015661707126076742)\n",
      "('liv.39.txt', 14682, 23, 0.0015665440675657267)\n",
      "('34.txt', 637, 1, 0.0015698587127158557)\n",
      "('serm13.txt', 1910, 3, 0.0015706806282722514)\n",
      "('5.txt', 9518, 15, 0.0015759613364152133)\n",
      "('liv.37.txt', 16390, 26, 0.0015863331299572911)\n",
      "('lucretius6.txt', 8815, 14, 0.0015882019285309132)\n",
      "('tac.ger.txt', 5665, 9, 0.001588702559576346)\n",
      "('inventione1.txt', 14964, 24, 0.0016038492381716118)\n",
      "('verres.2.5.txt', 19272, 31, 0.0016085512660855127)\n",
      "('serm10.txt', 2484, 4, 0.001610305958132045)\n",
      "('adbrutum2.txt', 1861, 3, 0.0016120365394948952)\n",
      "('liv.30.txt', 13632, 22, 0.001613849765258216)\n",
      "('seneca.ep19.txt', 8674, 14, 0.0016140189070786258)\n",
      "('suet.caesar.txt', 9862, 16, 0.0016223889677550193)\n",
      "('rutiliuslupus.txt', 4313, 7, 0.001623000231857176)\n",
      "('serm72.txt', 1231, 2, 0.0016246953696181965)\n",
      "('quintilian.decl.mai5.txt', 4912, 8, 0.0016286644951140066)\n",
      "('balbo.txt', 6725, 11, 0.0016356877323420074)\n",
      "('malachias.txt', 1220, 2, 0.001639344262295082)\n",
      "('varia5.txt', 9752, 16, 0.0016406890894175555)\n",
      "('thesauro.txt', 1824, 3, 0.001644736842105263)\n",
      "('malaterra1.txt', 7814, 13, 0.0016636805733299207)\n",
      "('44.txt', 1202, 2, 0.0016638935108153079)\n",
      "('pliny.ep3.txt', 6594, 11, 0.0016681831968456172)\n",
      "('varia12.txt', 8964, 15, 0.0016733601070950468)\n",
      "('tac.hist2.txt', 12366, 21, 0.001698204754973314)\n",
      "('phil8.txt', 2937, 5, 0.0017024174327545114)\n",
      "('tertullian.paen.txt', 4101, 7, 0.001706900755913192)\n",
      "('phil6.txt', 1755, 3, 0.0017094017094017094)\n",
      "('varia4.txt', 8188, 14, 0.001709819247679531)\n",
      "('capellanus1.txt', 32653, 56, 0.0017150032156310294)\n",
      "('fam10.txt', 10434, 18, 0.0017251293847038527)\n",
      "('sex.rosc.txt', 13309, 23, 0.0017281538808325194)\n",
      "('lucretius5.txt', 9806, 17, 0.0017336324699163777)\n",
      "('trin1.txt', 10953, 19, 0.0017346845613074044)\n",
      "('demort.txt', 10923, 19, 0.001739448869358235)\n",
      "('sen.qn7.txt', 6315, 11, 0.0017418844022169437)\n",
      "('trin14.txt', 8577, 15, 0.0017488632388947185)\n",
      "('histrom2.txt', 2830, 5, 0.0017667844522968198)\n",
      "('liv.40.txt', 14714, 26, 0.0017670246024194644)\n",
      "('fam5.txt', 8460, 15, 0.0017730496453900709)\n",
      "('quintilian.decl.mai8.txt', 4490, 8, 0.0017817371937639199)\n",
      "('tertullian.castitatis.txt', 3915, 7, 0.0017879948914431673)\n",
      "('mirabilia1.txt', 4468, 8, 0.0017905102954341987)\n",
      "('quintilian.decl.mai18.txt', 3349, 6, 0.0017915795759928337)\n",
      "('milo.txt', 10544, 19, 0.0018019726858877086)\n",
      "('nithardus2.txt', 3864, 7, 0.0018115942028985507)\n",
      "('tac.dialogus.txt', 9355, 17, 0.001817210048102619)\n",
      "('plancio.txt', 11492, 21, 0.0018273581621997912)\n",
      "('compet.txt', 4371, 8, 0.0018302447952413636)\n",
      "('quintilian.decl.mai13.txt', 3811, 7, 0.0018367882445552348)\n",
      "('lucretius3.txt', 7619, 14, 0.0018375114844467778)\n",
      "('serm15.txt', 2176, 4, 0.001838235294117647)\n",
      "('nepos.cim.txt', 544, 1, 0.001838235294117647)\n",
      "('lig.txt', 3256, 6, 0.0018427518427518428)\n",
      "('quintilian.decl.mai14.txt', 2166, 4, 0.0018467220683287165)\n",
      "('off2.txt', 8644, 16, 0.0018509949097639982)\n",
      "('solet.txt', 1615, 3, 0.0018575851393188853)\n",
      "('phil2.txt', 11299, 21, 0.0018585715550048676)\n",
      "('columella.rr3.txt', 10195, 19, 0.0018636586562040215)\n",
      "('bc2.txt', 6435, 12, 0.0018648018648018648)\n",
      "('phil3.txt', 3750, 7, 0.0018666666666666666)\n",
      "('eutropius2.txt', 2132, 4, 0.001876172607879925)\n",
      "('lucretius1.txt', 7450, 14, 0.0018791946308724832)\n",
      "('tac.ann5.txt', 531, 1, 0.0018832391713747645)\n",
      "('ambro.txt', 530, 1, 0.0018867924528301887)\n",
      "('ben2.txt', 6347, 12, 0.0018906570033086498)\n",
      "('liv.33.txt', 11631, 22, 0.0018914968618347519)\n",
      "('asconius.txt', 15857, 30, 0.0018919089361165416)\n",
      "('cael.txt', 8434, 16, 0.001897083234526915)\n",
      "('nepos.ham.txt', 523, 1, 0.0019120458891013384)\n",
      "('fam12.txt', 8361, 16, 0.0019136466929793087)\n",
      "('orosius6.txt', 11994, 23, 0.00191762547940637)\n",
      "('sen.ira3.txt', 9364, 18, 0.0019222554463904314)\n",
      "('poggio.txt', 14023, 27, 0.0019254082578620837)\n",
      "('senectute.txt', 8259, 16, 0.001937280542438552)\n",
      "('institutes2.txt', 17011, 33, 0.0019399212274410675)\n",
      "('tac.ann13.txt', 7716, 15, 0.0019440124416796269)\n",
      "('institutes1.txt', 8213, 16, 0.0019481310118105442)\n",
      "('sall.ep1.txt', 2553, 5, 0.0019584802193497847)\n",
      "('oratore2.txt', 27054, 53, 0.00195904487321653)\n",
      "('trin12.txt', 5594, 11, 0.001966392563460851)\n",
      "('curtius10.txt', 6061, 12, 0.001979871308364956)\n",
      "('ben1.txt', 4009, 8, 0.001995510102269893)\n",
      "('serm9.txt', 7009, 14, 0.00199743187330575)\n",
      "('43.txt', 1493, 3, 0.0020093770931011385)\n",
      "('orosius7.txt', 17389, 35, 0.002012766691586635)\n",
      "('mariamag.txt', 992, 2, 0.0020161290322580645)\n",
      "('lucretius4.txt', 8904, 18, 0.0020215633423180594)\n",
      "('15.txt', 10801, 22, 0.002036848439959263)\n",
      "('sall.ep2.txt', 1471, 3, 0.0020394289598912306)\n",
      "('trin9.txt', 4896, 10, 0.002042483660130719)\n",
      "('serm88.txt', 5807, 12, 0.002066471499913897)\n",
      "('sen.consolatione3.txt', 6769, 14, 0.002068252326783868)\n",
      "('apuleius10.txt', 5794, 12, 0.0020711080428028996)\n",
      "('bacon.intro.txt', 481, 1, 0.002079002079002079)\n",
      "('jonah.txt', 961, 2, 0.002081165452653486)\n",
      "('ennodius.txt', 4775, 10, 0.0020942408376963353)\n",
      "('columella.rr1.txt', 8107, 17, 0.0020969532502775378)\n",
      "('sen.brevita.txt', 6164, 13, 0.002109020116807268)\n",
      "('ovid.her19.txt', 1420, 3, 0.002112676056338028)\n",
      "('off1.txt', 14642, 31, 0.002117197104220735)\n",
      "('tac.ann12.txt', 7552, 16, 0.00211864406779661)\n",
      "('varia3.txt', 9889, 21, 0.002123571645262413)\n",
      "('hist5.txt', 6116, 13, 0.0021255722694571615)\n",
      "('verres.2.3.txt', 24880, 53, 0.002130225080385852)\n",
      "('nepos.age.txt', 1406, 3, 0.0021337126600284497)\n",
      "('hist4.txt', 6559, 14, 0.0021344717182497333)\n",
      "('vegetius2.txt', 4665, 10, 0.0021436227224008574)\n",
      "('sen.consolatione2.txt', 8383, 18, 0.0021472026720744364)\n",
      "('liv.21.txt', 15827, 34, 0.0021482277121374865)\n",
      "('nepos.dat.txt', 1852, 4, 0.0021598272138228943)\n",
      "('quintilian.decl.mai11.txt', 2315, 5, 0.0021598272138228943)\n",
      "('apuleius6.txt', 4624, 10, 0.0021626297577854673)\n",
      "('sen.qn3.txt', 7834, 17, 0.0021700280827163646)\n",
      "('caesar.txt', 14679, 32, 0.002179985012603038)\n",
      "('4.txt', 10087, 22, 0.0021810250817884407)\n",
      "('vell2.txt', 22839, 50, 0.002189237707430273)\n",
      "('38.txt', 2740, 6, 0.0021897810218978104)\n",
      "('phil11.txt', 4096, 9, 0.002197265625)\n",
      "('nithardus4.txt', 2728, 6, 0.0021994134897360706)\n",
      "('columella.rr2.txt', 10408, 23, 0.0022098385857033053)\n",
      "('gestarom.txt', 15827, 35, 0.0022114108801415304)\n",
      "('sall.phil.txt', 904, 2, 0.0022123893805309734)\n",
      "('vegetius3.txt', 9943, 22, 0.002212611887760233)\n",
      "('rabiriopost.txt', 4061, 9, 0.002216202905688254)\n",
      "('seneca.ep3.txt', 5409, 12, 0.0022185246810870773)\n",
      "('max.txt', 5389, 12, 0.002226758211170904)\n",
      "('quintilian.institutio4.txt', 12061, 27, 0.0022386203465715945)\n",
      "('sen.qn5.txt', 3995, 9, 0.0022528160200250315)\n",
      "('sall.cotta.txt', 443, 1, 0.002257336343115124)\n",
      "('sen.constantia.txt', 5312, 12, 0.002259036144578313)\n",
      "('liv.34.txt', 15035, 34, 0.0022613900897904887)\n",
      "('seneca.contr3.txt', 3093, 7, 0.0022631749110895573)\n",
      "('arnobius1.txt', 10601, 24, 0.002263937364399585)\n",
      "('valmax2.txt', 8811, 20, 0.0022698899103393484)\n",
      "('passerat.txt', 440, 1, 0.0022727272727272726)\n",
      "('nepos.cat.txt', 439, 1, 0.002277904328018223)\n",
      "('phil7.txt', 2193, 5, 0.002279981760145919)\n",
      "('seneca.contr8.txt', 1749, 4, 0.002287021154945683)\n",
      "('phil13.txt', 5232, 12, 0.0022935779816513763)\n",
      "('victor.caes.txt', 10859, 25, 0.002302237775117414)\n",
      "('sen.ira1.txt', 5609, 13, 0.002317703690497415)\n",
      "('bacon.praefatio.txt', 2585, 6, 0.0023210831721470018)\n",
      "('orosius2.txt', 6885, 16, 0.0023238925199709514)\n",
      "('7.txt', 10262, 24, 0.0023387253946599105)\n",
      "('gaius3.txt', 11096, 26, 0.002343186733958183)\n",
      "('seneca.contr5.txt', 1702, 4, 0.0023501762632197414)\n",
      "('curtius3.txt', 7205, 17, 0.002359472588480222)\n",
      "('tertullian.uxor2.txt', 2117, 5, 0.0023618327822390174)\n",
      "('gestafrancorum1.txt', 1683, 4, 0.0023767082590612004)\n",
      "('rhetorica.txt', 10096, 24, 0.002377179080824089)\n",
      "('valmax4.txt', 9240, 22, 0.002380952380952381)\n",
      "('histrom4.txt', 2938, 7, 0.002382573179033356)\n",
      "('wisdom.txt', 7135, 17, 0.0023826208829712684)\n",
      "('2.txt', 8793, 21, 0.002388263391334016)\n",
      "('pascha.txt', 1253, 3, 0.0023942537909018356)\n",
      "('sall.pomp.txt', 415, 1, 0.0024096385542168677)\n",
      "('liv.38.txt', 16992, 41, 0.002412900188323917)\n",
      "('curtius6.txt', 8679, 21, 0.0024196335983408227)\n",
      "('chris.txt', 826, 2, 0.002421307506053269)\n",
      "('fam1.txt', 7420, 18, 0.002425876010781671)\n",
      "('histrom12.txt', 2060, 5, 0.0024271844660194173)\n",
      "('nepos.iph.txt', 411, 1, 0.0024330900243309003)\n",
      "('vitamalchus.txt', 2050, 5, 0.0024390243902439024)\n",
      "('geta.txt', 1229, 3, 0.0024410089503661514)\n",
      "('florus1.txt', 16788, 41, 0.0024422206337860377)\n",
      "('quintilian.institutio7.txt', 13070, 32, 0.002448355011476664)\n",
      "('pascoli.sen.txt', 1224, 3, 0.0024509803921568627)\n",
      "('suet.terence.txt', 816, 2, 0.0024509803921568627)\n",
      "('oratore1.txt', 18341, 45, 0.002453519437326209)\n",
      "('arnobius7.txt', 12199, 30, 0.002459217968685958)\n",
      "('seneca.ep4.txt', 5689, 14, 0.0024608894357532078)\n",
      "('sen.qn2.txt', 8513, 21, 0.002466815458710208)\n",
      "('prud4.txt', 806, 2, 0.0024813895781637717)\n",
      "('flacco.txt', 10856, 27, 0.0024871039056742813)\n",
      "('columella.rr4.txt', 9241, 23, 0.002488908126826101)\n",
      "('fam13.txt', 12850, 32, 0.002490272373540856)\n",
      "('civ3.txt', 10837, 27, 0.002491464427424564)\n",
      "('leg1.txt', 6421, 16, 0.0024918237034729793)\n",
      "('car.txt', 2004, 5, 0.00249500998003992)\n",
      "('41.txt', 1202, 3, 0.0024958402662229617)\n",
      "('valmax5.txt', 10014, 25, 0.0024965048931495904)\n",
      "('bacon.praefatio2.txt', 1201, 3, 0.002497918401332223)\n",
      "('tertullian.virginibus.txt', 5603, 14, 0.0024986614313760487)\n",
      "('apuleius.cupid.txt', 9157, 23, 0.0025117396527246914)\n",
      "('seneca.ep9.txt', 7148, 18, 0.002518186905428092)\n",
      "('fin3.txt', 7931, 20, 0.0025217500945656285)\n",
      "('sen.qn4.txt', 5150, 13, 0.002524271844660194)\n",
      "('pliny.ep10.txt', 9463, 24, 0.00253619359611117)\n",
      "('sen.otio.txt', 1969, 5, 0.0025393600812595226)\n",
      "('sen.qn1.txt', 7858, 20, 0.0025451768897938407)\n",
      "('inventione2.txt', 16889, 43, 0.0025460358813428857)\n",
      "('29.txt', 784, 2, 0.002551020408163265)\n",
      "('apuleius5.txt', 4704, 12, 0.002551020408163265)\n",
      "('rosccom.txt', 4696, 12, 0.002555366269165247)\n",
      "('maximianus.txt', 4298, 11, 0.0025593299208934387)\n",
      "('phil12.txt', 3117, 8, 0.0025665704202759063)\n",
      "('orosius4.txt', 10495, 27, 0.0025726536445926634)\n",
      "('florus2.txt', 10097, 26, 0.002575022283846687)\n",
      "('tac.ann2.txt', 10461, 27, 0.0025810151993117293)\n",
      "('valesianus1.txt', 1549, 4, 0.0025823111684958036)\n",
      "('tac.hist5.txt', 3481, 9, 0.002585463947141626)\n",
      "('sen.qn6.txt', 8115, 21, 0.002587800369685767)\n",
      "('civ4.txt', 10389, 27, 0.0025989026855327752)\n",
      "('ente.txt', 7270, 19, 0.0026134800550206326)\n",
      "('prud13.txt', 765, 2, 0.00261437908496732)\n",
      "('quintilian.decl.mai17.txt', 3819, 10, 0.002618486514794449)\n",
      "('ecclesiastes.txt', 3811, 10, 0.0026239832065074785)\n",
      "('greg.txt', 762, 2, 0.0026246719160104987)\n",
      "('3.txt', 9135, 24, 0.002627257799671593)\n",
      "('18.txt', 14841, 39, 0.0026278552658176675)\n",
      "('suet.titus.txt', 1522, 4, 0.002628120893561104)\n",
      "('liv.36.txt', 11411, 30, 0.0026290421523091755)\n",
      "('serm20.txt', 1520, 4, 0.002631578947368421)\n",
      "('quintilian.decl.mai10.txt', 3787, 10, 0.0026406126221283337)\n",
      "('valmax6.txt', 8710, 23, 0.002640642939150402)\n",
      "('serm87.txt', 3407, 9, 0.0026416201937188143)\n",
      "('conf11.txt', 6804, 18, 0.0026455026455026454)\n",
      "('q1.49.txt', 2266, 6, 0.00264783759929391)\n",
      "('liv.24.txt', 14348, 38, 0.002648452746027321)\n",
      "('apuleius4.txt', 5258, 14, 0.0026626093571700264)\n",
      "('33.txt', 375, 1, 0.0026666666666666666)\n",
      "('trin4.txt', 9746, 26, 0.0026677611327724196)\n",
      "('epistula.txt', 374, 1, 0.00267379679144385)\n",
      "('serm83.txt', 1868, 5, 0.0026766595289079227)\n",
      "('sidonius5.txt', 4481, 12, 0.0026779736665922785)\n",
      "('fratrem1.txt', 7817, 21, 0.0026864526033004987)\n",
      "('nd2.txt', 15624, 42, 0.002688172043010753)\n",
      "('aelredus.txt', 16706, 45, 0.002693643002514067)\n",
      "('varia2.txt', 8904, 24, 0.0026954177897574125)\n",
      "('falcone.txt', 34445, 93, 0.0026999564523152854)\n",
      "('32.txt', 1111, 3, 0.0027002700270027003)\n",
      "('seneca.ep7.txt', 7398, 20, 0.002703433360367667)\n",
      "('marcant.txt', 5547, 15, 0.0027041644131963224)\n",
      "('apuleius9.txt', 6655, 18, 0.00270473328324568)\n",
      "('repub6.txt', 2216, 6, 0.002707581227436823)\n",
      "('varia1.txt', 10331, 28, 0.002710289420191656)\n",
      "('seneca.ep1.txt', 6252, 17, 0.0027191298784388996)\n",
      "('liv.31.txt', 12815, 35, 0.0027311744049941474)\n",
      "('liv.44.txt', 12801, 35, 0.0027341613936411216)\n",
      "('nepos.fragmenta.txt', 365, 1, 0.0027397260273972603)\n",
      "('serm81.txt', 2911, 8, 0.0027481964960494676)\n",
      "('fato.txt', 5090, 14, 0.00275049115913556)\n",
      "('gestafrancorum5.txt', 727, 2, 0.002751031636863824)\n",
      "('sen.tranq.txt', 7602, 21, 0.0027624309392265192)\n",
      "('sidonius6.txt', 2533, 7, 0.0027635215159889457)\n",
      "('valmax7.txt', 7580, 21, 0.002770448548812665)\n",
      "('formula.txt', 2162, 6, 0.0027752081406105457)\n",
      "('apuleius11.txt', 5034, 14, 0.0027810885975367503)\n",
      "('6.txt', 10026, 28, 0.0027927388789148215)\n",
      "('sen.proverbs.txt', 1428, 4, 0.0028011204481792717)\n",
      "('capellanus3.txt', 6769, 19, 0.0028069138720638203)\n",
      "('obsequens.txt', 4984, 14, 0.0028089887640449437)\n",
      "('georgio.txt', 712, 2, 0.0028089887640449437)\n",
      "('liv.43.txt', 5693, 16, 0.002810468997013877)\n",
      "('fam7.txt', 7461, 21, 0.0028146361077603537)\n",
      "('hist6.txt', 6743, 19, 0.002817736912353552)\n",
      "('quintilian.decl.mai6.txt', 4968, 14, 0.002818035426731079)\n",
      "('12.txt', 9569, 27, 0.0028216114536524194)\n",
      "('anima.txt', 10257, 29, 0.0028273374280978845)\n",
      "('clitophon.txt', 1412, 4, 0.0028328611898017)\n",
      "('fin2.txt', 12657, 36, 0.002844275894761792)\n",
      "('17.txt', 12938, 37, 0.002859792858247024)\n",
      "('quintilian.institutio12.txt', 13268, 38, 0.0028640337654507087)\n",
      "('quintilian.institutio10.txt', 12207, 35, 0.0028672073400507906)\n",
      "('seneca.ep2.txt', 6624, 19, 0.0028683574879227055)\n",
      "('rabirio.txt', 3485, 10, 0.0028694404591104736)\n",
      "('sen.vita.txt', 7307, 21, 0.0028739564800875873)\n",
      "('apuleius2.txt', 4867, 14, 0.0028765153071707416)\n",
      "('serm14.txt', 1738, 5, 0.0028768699654775605)\n",
      "('gestafrancorum9.txt', 4859, 14, 0.0028812512862728956)\n",
      "('suet.vit.txt', 2426, 7, 0.0028854080791426216)\n",
      "('tac.ann15.txt', 9345, 27, 0.0028892455858747996)\n",
      "('celtis.odes.txt', 346, 1, 0.002890173410404624)\n",
      "('gord.txt', 5512, 16, 0.002902757619738752)\n",
      "('oratore3.txt', 16854, 49, 0.0029073217040465173)\n",
      "('fin1.txt', 7204, 21, 0.002915047196002221)\n",
      "('silv.txt', 1368, 4, 0.0029239766081871343)\n",
      "('piso.txt', 10933, 32, 0.002926918503612915)\n",
      "('37.txt', 683, 2, 0.0029282576866764276)\n",
      "('hugo5.txt', 3753, 11, 0.002930988542499334)\n",
      "('histrom13.txt', 1705, 5, 0.002932551319648094)\n",
      "('leg3.txt', 4773, 14, 0.002933165723863398)\n",
      "('pliny.ep8.txt', 6132, 18, 0.0029354207436399216)\n",
      "('curtius9.txt', 8164, 24, 0.002939735423811857)\n",
      "('serm8.txt', 4751, 14, 0.002946748053041465)\n",
      "('columella.rr5.txt', 9162, 27, 0.0029469548133595285)\n",
      "('gravissimas.txt', 1692, 5, 0.002955082742316785)\n",
      "('richerus4.txt', 15753, 47, 0.0029835586872341774)\n",
      "('16.txt', 12381, 37, 0.002988450044422906)\n",
      "('pascoli.veianius.txt', 669, 2, 0.0029895366218236174)\n",
      "('diad.txt', 1669, 5, 0.0029958058717795086)\n",
      "('tertullian.monog.txt', 6959, 21, 0.003017674953297888)\n",
      "('seneca.contr4.txt', 2981, 9, 0.0030191211003019122)\n",
      "('valmax1.txt', 8265, 25, 0.0030248033877797943)\n",
      "('solinus1.txt', 8263, 25, 0.0030255355197870023)\n",
      "('sidonius2.txt', 4957, 15, 0.003026023804720597)\n",
      "('35.txt', 330, 1, 0.0030303030303030303)\n",
      "('serm92.txt', 658, 2, 0.00303951367781155)\n",
      "('esdras2.txt', 15725, 48, 0.0030524642289348172)\n",
      "('tac.ann14.txt', 8187, 25, 0.0030536215952119213)\n",
      "('bacon.hist3.txt', 6218, 19, 0.0030556449018977164)\n",
      "('21.txt', 12415, 38, 0.0030608135320177206)\n",
      "('brut.txt', 24820, 76, 0.0030620467365028202)\n",
      "('leviticus.txt', 14030, 43, 0.0030648610121168924)\n",
      "('bacon.epistola.txt', 326, 1, 0.003067484662576687)\n",
      "('gestafrancorum6.txt', 1302, 4, 0.0030721966205837174)\n",
      "('epistulae.txt', 5184, 16, 0.0030864197530864196)\n",
      "('pert.txt', 2590, 8, 0.003088803088803089)\n",
      "('jud.txt', 646, 2, 0.0030959752321981426)\n",
      "('q1.60.txt', 2904, 9, 0.0030991735537190084)\n",
      "('malaterra3.txt', 9335, 29, 0.0031065881092662024)\n",
      "('22.txt', 15734, 49, 0.0031142748188636073)\n",
      "('sidonius3.txt', 3853, 12, 0.003114456267843239)\n",
      "('civ6.txt', 7371, 23, 0.003120336453669787)\n",
      "('columella.rr11.txt', 10885, 34, 0.003123564538355535)\n",
      "('att11.txt', 7033, 22, 0.0031281103369827953)\n",
      "('liv.per7.txt', 319, 1, 0.003134796238244514)\n",
      "('waltarius1.txt', 2870, 9, 0.0031358885017421603)\n",
      "('vitruvius8.txt', 5738, 18, 0.003136981526664343)\n",
      "('fam3.txt', 7006, 22, 0.003140165572366543)\n",
      "('seneca.contr7.txt', 16231, 51, 0.003142135419875547)\n",
      "('buchanan.txt', 5086, 16, 0.0031458906802988595)\n",
      "('repub1.txt', 8230, 26, 0.0031591737545565005)\n",
      "('13.txt', 11697, 37, 0.0031632042404035224)\n",
      "('23.txt', 947, 3, 0.0031678986272439284)\n",
      "('macabees2.txt', 10414, 33, 0.003168811215671212)\n",
      "('malaterra4.txt', 9144, 29, 0.0031714785651793527)\n",
      "('curtius4.txt', 12865, 41, 0.0031869413136416634)\n",
      "('victor.caes2.txt', 9093, 29, 0.0031892664687121963)\n",
      "('silius15.txt', 5315, 17, 0.003198494825964252)\n",
      "('fam2.txt', 5627, 18, 0.0031988626266216454)\n",
      "('histrom6.txt', 2810, 9, 0.003202846975088968)\n",
      "('serm71.txt', 8734, 28, 0.003205862147927639)\n",
      "('serm80.txt', 2182, 7, 0.0032080659945004585)\n",
      "('richerus1.txt', 8101, 26, 0.003209480311072707)\n",
      "('sidonius7.txt', 7141, 23, 0.0032208374177286093)\n",
      "('anon.martyrio.txt', 2794, 9, 0.0032211882605583395)\n",
      "('30.txt', 6518, 21, 0.0032218471923903037)\n",
      "('civ1.txt', 11132, 36, 0.0032339202299676607)\n",
      "('histrom11.txt', 1546, 5, 0.003234152652005175)\n",
      "('quintilian.institutio5.txt', 9541, 31, 0.003249135310764071)\n",
      "('desummobono.txt', 2149, 7, 0.003257328990228013)\n",
      "('liv.42.txt', 16858, 55, 0.0032625459722386996)\n",
      "('repub2.txt', 6114, 20, 0.0032711808963035655)\n",
      "('sulpiciusseverusmartin.txt', 6725, 22, 0.0032713754646840148)\n",
      "('fam11.txt', 6722, 22, 0.0032728354656352274)\n",
      "('fonteio.txt', 4569, 15, 0.003282994090610637)\n",
      "('orosius3.txt', 8811, 29, 0.0032913403699920555)\n",
      "('fin4.txt', 8507, 28, 0.0032914070765252147)\n",
      "('phaedr3.txt', 2418, 8, 0.0033085194375516956)\n",
      "('dia.txt', 5735, 19, 0.0033129904097646034)\n",
      "('liv.45.txt', 13274, 44, 0.0033147506403495555)\n",
      "('sepsev.txt', 4199, 14, 0.003334127173136461)\n",
      "('propos.txt', 1798, 6, 0.0033370411568409346)\n",
      "('tusc5.txt', 11071, 37, 0.0033420648541233854)\n",
      "('28.txt', 895, 3, 0.0033519553072625698)\n",
      "('anselmproslogion.txt', 6259, 21, 0.0033551685572775205)\n",
      "('marina.txt', 298, 1, 0.003355704697986577)\n",
      "('q1.7.txt', 2376, 8, 0.003367003367003367)\n",
      "('psplato.demodocus.txt', 2375, 8, 0.003368421052631579)\n",
      "('valesianus2.txt', 2961, 10, 0.003377237419790611)\n",
      "('q1.82.txt', 2960, 10, 0.0033783783783783786)\n",
      "('hugo.solo.txt', 7987, 27, 0.0033804933016151244)\n",
      "('seneca.contr2.txt', 15676, 53, 0.0033809645317683084)\n",
      "('tertullian.adsenatorem.txt', 591, 2, 0.00338409475465313)\n",
      "('tertullian.marcionem1.txt', 10333, 35, 0.0033872060389044806)\n",
      "('aqua1.txt', 4104, 14, 0.00341130604288499)\n",
      "('off3.txt', 11135, 38, 0.0034126627750336778)\n",
      "('25.txt', 878, 3, 0.003416856492027335)\n",
      "('ezekiel.txt', 26853, 92, 0.003426060402934495)\n",
      "('hyginus6.txt', 2914, 10, 0.0034317089910775567)\n",
      "('prud11.txt', 1453, 5, 0.0034411562284927736)\n",
      "('richerus2.txt', 13655, 47, 0.003441962651043574)\n",
      "('divinatione2.txt', 14771, 51, 0.0034527113939476)\n",
      "('apuleius8.txt', 4914, 17, 0.0034595034595034596)\n",
      "('seneca.ep11-13.txt', 8938, 31, 0.0034683374356679346)\n",
      "('aelii.txt', 1441, 5, 0.0034698126301179735)\n",
      "('tertullian.marcionem2.txt', 10349, 36, 0.003478596965890424)\n",
      "('histapoll.txt', 11669, 41, 0.003513582997686177)\n",
      "('gestafrancorum7.txt', 851, 3, 0.0035252643948296123)\n",
      "('tertullian.resurrectione.txt', 22645, 80, 0.0035327886950761757)\n",
      "('varro.ll9.txt', 7356, 26, 0.0035345296356715608)\n",
      "('suet.otho.txt', 1694, 6, 0.0035419126328217238)\n",
      "('20.txt', 12696, 45, 0.003544423440453686)\n",
      "('more.txt', 27629, 98, 0.00354699771978718)\n",
      "('gall.txt', 3644, 13, 0.0035675082327113064)\n",
      "('arnobius2.txt', 15114, 54, 0.003572846367606193)\n",
      "('divinatione1.txt', 12867, 46, 0.0035750369161420686)\n",
      "('tertullian.fuga.txt', 5308, 19, 0.0035795026375282594)\n",
      "('nepos.tim.txt', 837, 3, 0.0035842293906810036)\n",
      "('serm90.txt', 3333, 12, 0.0036003600360036)\n",
      "('tusc2.txt', 6655, 24, 0.0036063110443275734)\n",
      "('9.txt', 8041, 29, 0.003606516602412635)\n",
      "('piccolomini.turcos.txt', 1663, 6, 0.0036079374624173183)\n",
      "('conf10.txt', 11640, 42, 0.0036082474226804126)\n",
      "('14.txt', 11622, 42, 0.0036138358286009293)\n",
      "('marx.txt', 830, 3, 0.0036144578313253013)\n",
      "('marc.txt', 552, 2, 0.0036231884057971015)\n",
      "('probus.txt', 4138, 15, 0.003624939584340261)\n",
      "('seneca.contr9.txt', 13225, 48, 0.003629489603024575)\n",
      "('seneca.ep10.txt', 4959, 18, 0.003629764065335753)\n",
      "('conf9.txt', 6057, 22, 0.0036321611358758462)\n",
      "('columella.rr7.txt', 7425, 27, 0.0036363636363636364)\n",
      "('deaeternitate.txt', 7680, 28, 0.0036458333333333334)\n",
      "('orator.txt', 18630, 68, 0.0036500268384326356)\n",
      "('gregorytours3.txt', 7942, 29, 0.003651473180559053)\n",
      "('bede1.txt', 13691, 50, 0.003652034183039953)\n",
      "('donation.txt', 2725, 10, 0.003669724770642202)\n",
      "('civ5.txt', 12256, 45, 0.0036716710182767625)\n",
      "('arnobius6.txt', 5719, 21, 0.0036719706242350062)\n",
      "('vin.txt', 1089, 4, 0.0036730945821854912)\n",
      "('q1.37.txt', 1902, 7, 0.003680336487907466)\n",
      "('joel.txt', 1358, 5, 0.003681885125184094)\n",
      "('serm82.txt', 3258, 12, 0.003683241252302026)\n",
      "('sidonius4.txt', 7595, 28, 0.003686635944700461)\n",
      "('ein.txt', 7591, 28, 0.0036885785798972466)\n",
      "('fredegarius.txt', 21936, 81, 0.003692560175054705)\n",
      "('valmax8.txt', 8666, 32, 0.0036925917378259868)\n",
      "('conf8.txt', 5682, 21, 0.003695881731784583)\n",
      "('silius10.txt', 4328, 16, 0.0036968576709796672)\n",
      "('clod.txt', 2695, 10, 0.0037105751391465678)\n",
      "('hadr.txt', 5120, 19, 0.0037109375)\n",
      "('serm11.txt', 808, 3, 0.0037128712871287127)\n",
      "('apuleius1.txt', 4039, 15, 0.0037137905422134194)\n",
      "('tusc4.txt', 8332, 31, 0.0037205952952472396)\n",
      "('lucan1.txt', 4565, 17, 0.003723986856516977)\n",
      "('lucan4.txt', 5359, 20, 0.003732039559619332)\n",
      "('sidonius1.txt', 5622, 21, 0.003735325506937033)\n",
      "('prud1.txt', 803, 3, 0.0037359900373599006)\n",
      "('zephaniah.txt', 1069, 4, 0.0037418147801683817)\n",
      "('prud10.txt', 6144, 23, 0.0037434895833333335)\n",
      "('arnobius5.txt', 9072, 34, 0.003747795414462081)\n",
      "('exivi.txt', 3997, 15, 0.0037528146109582186)\n",
      "('reg.txt', 2127, 8, 0.0037611659614480487)\n",
      "('sidonius8.txt', 6379, 24, 0.003762345195171657)\n",
      "('hipp.txt', 2123, 8, 0.0037682524729156855)\n",
      "('seneca.contr1.txt', 17123, 65, 0.003796063773871401)\n",
      "('hist12.txt', 7899, 30, 0.0037979491074819596)\n",
      "('civ20.txt', 20000, 76, 0.0038)\n",
      "('seneca.ep14-15.txt', 16822, 64, 0.003804541671620497)\n",
      "('liv.32.txt', 10766, 41, 0.003808285342745681)\n",
      "('11.txt', 13914, 53, 0.0038091131234727614)\n",
      "('liv.per9.txt', 262, 1, 0.003816793893129771)\n",
      "('cato.dis.txt', 2095, 8, 0.0038186157517899762)\n",
      "('optgen.txt', 1569, 6, 0.0038240917782026767)\n",
      "('trin5.txt', 5226, 20, 0.003827018752391887)\n",
      "('job.txt', 12540, 48, 0.003827751196172249)\n",
      "('suet.tib.txt', 9386, 36, 0.0038354996803750267)\n",
      "('proverbs.txt', 9892, 38, 0.003841488071168621)\n",
      "('arnobius3.txt', 6753, 26, 0.0038501406782170886)\n",
      "('prud2.txt', 2076, 8, 0.0038535645472061657)\n",
      "('iul.txt', 519, 2, 0.0038535645472061657)\n",
      "('bacon.hist10.txt', 3629, 14, 0.0038578120694406173)\n",
      "('micah.txt', 2073, 8, 0.003859141341051616)\n",
      "('benedict.txt', 13198, 51, 0.0038642218517957265)\n",
      "('liv.35.txt', 12675, 49, 0.003865877712031558)\n",
      "('isaiah.txt', 24572, 95, 0.003866189158391665)\n",
      "('minucius.txt', 11635, 45, 0.003867640739149119)\n",
      "('resgestae.txt', 2584, 10, 0.003869969040247678)\n",
      "('nepos.pr.txt', 258, 1, 0.003875968992248062)\n",
      "('legagr1.txt', 2318, 9, 0.003882657463330457)\n",
      "('zacharias.txt', 4378, 17, 0.0038830516217450892)\n",
      "('19.txt', 12088, 47, 0.0038881535407015223)\n",
      "('liv.per3.txt', 257, 1, 0.0038910505836575876)\n",
      "('phaedrapp.txt', 2564, 10, 0.0039001560062402497)\n",
      "('lucan2.txt', 4871, 19, 0.00390063641962636)\n",
      "('seneca.ep5.txt', 6387, 25, 0.003914200720212932)\n",
      "('capitula.txt', 4339, 17, 0.003917953445494353)\n",
      "('luther.lteramus.txt', 762, 3, 0.003937007874015748)\n",
      "('columella.rr8.txt', 8622, 34, 0.003943400603108328)\n",
      "('curtius5.txt', 8102, 32, 0.003949642063687978)\n",
      "('sirach.txt', 20474, 81, 0.003956237178860994)\n",
      "('serm75.txt', 1768, 7, 0.003959276018099547)\n",
      "('habakkuk.txt', 1010, 4, 0.0039603960396039604)\n",
      "('inst.txt', 26711, 106, 0.003968402530792558)\n",
      "('fam8.txt', 6293, 25, 0.003972668043858255)\n",
      "('princ.txt', 1507, 6, 0.003981420039814201)\n",
      "('39.txt', 1251, 5, 0.003996802557953637)\n",
      "('mark.txt', 10250, 41, 0.004)\n",
      "('suet.aug.txt', 13974, 56, 0.0040074423930156)\n",
      "('conf7.txt', 5976, 24, 0.004016064257028112)\n",
      "('vitruvius2.txt', 6967, 28, 0.0040189464618917755)\n",
      "('q1.36.txt', 3976, 16, 0.004024144869215292)\n",
      "('querela.txt', 9924, 40, 0.004030632809351068)\n",
      "('eugippius.txt', 11399, 46, 0.004035441705412756)\n",
      "('com.txt', 3468, 14, 0.004036908881199538)\n",
      "('theganus.txt', 6427, 26, 0.0040454333281468805)\n",
      "('civ15.txt', 14825, 60, 0.004047217537942664)\n",
      "('tertullian.praxean.txt', 14301, 58, 0.004055660443325642)\n",
      "('sall.mithr.txt', 738, 3, 0.0040650406504065045)\n",
      "('avienus.ora.txt', 3920, 16, 0.004081632653061225)\n",
      "('topica.txt', 6837, 28, 0.004095363463507386)\n",
      "('seneca.contr10.txt', 12207, 50, 0.004096010485786844)\n",
      "('acts.txt', 16573, 68, 0.004103059192662765)\n",
      "('hist3.txt', 5360, 22, 0.004104477611940298)\n",
      "('ovid.ponto1.txt', 5085, 21, 0.004129793510324484)\n",
      "('gwinne5.2.txt', 3627, 15, 0.0041356492969396195)\n",
      "('bacon.hist8.txt', 6043, 25, 0.004137018037398643)\n",
      "('antibarb.txt', 28246, 117, 0.004142179423635205)\n",
      "('silius16.txt', 4585, 19, 0.004143947655398037)\n",
      "('fam15.txt', 7234, 30, 0.004147083218136578)\n",
      "('8.txt', 10849, 45, 0.004147847727901189)\n",
      "('aurel.txt', 7703, 32, 0.004154225626379333)\n",
      "('fratrem2.txt', 5279, 22, 0.004167455957567721)\n",
      "('bacon.hist2.txt', 5512, 23, 0.0041727140783744555)\n",
      "('pliny.ep6.txt', 6704, 28, 0.004176610978520286)\n",
      "('sall.invectiva.txt', 714, 3, 0.004201680672268907)\n",
      "('psalms.txt', 30173, 127, 0.004209061081098996)\n",
      "('varro.rr1.txt', 14242, 60, 0.004212891447830361)\n",
      "('ovid.her20.txt', 1660, 7, 0.004216867469879518)\n",
      "('silius4.txt', 5448, 23, 0.004221732745961821)\n",
      "('avid.txt', 2594, 11, 0.004240555127216654)\n",
      "('sententiae.txt', 4478, 19, 0.004242965609647164)\n",
      "('ep.priapismo.txt', 2592, 11, 0.004243827160493827)\n",
      "('professio.txt', 1177, 5, 0.004248088360237893)\n",
      "('liv.per49.txt', 706, 3, 0.00424929178470255)\n",
      "('andreasbergoma.txt', 3995, 17, 0.00425531914893617)\n",
      "('seneca.contr6.txt', 1878, 8, 0.004259850905218318)\n",
      "('q1.81.txt', 1643, 7, 0.00426049908703591)\n",
      "('1.txt', 13842, 59, 0.004262389828059529)\n",
      "('pliny.ep2.txt', 6322, 27, 0.004270800379626701)\n",
      "('silius6.txt', 4670, 20, 0.004282655246252677)\n",
      "('arnobius4.txt', 6994, 30, 0.0042893909064912784)\n",
      "('curtius7.txt', 9558, 41, 0.004289600334798075)\n",
      "('henry1.txt', 3261, 14, 0.004293161606869059)\n",
      "('strena.txt', 6283, 27, 0.004297310202132739)\n",
      "('trin2.txt', 9767, 42, 0.004300194532609809)\n",
      "('solinus1a.txt', 9764, 42, 0.0043015157722244984)\n",
      "('suet.vesp.txt', 3254, 14, 0.0043023970497848806)\n",
      "('civ22.txt', 13000, 56, 0.0043076923076923075)\n",
      "('q1.35.txt', 928, 4, 0.004310344827586207)\n",
      "('gregorytours2.txt', 13448, 58, 0.004312908982748364)\n",
      "('apuleius3.txt', 4170, 18, 0.004316546762589928)\n",
      "('bacon.hist4.txt', 6482, 28, 0.004319654427645789)\n",
      "('bacon.hist7.txt', 7398, 32, 0.004325493376588267)\n",
      "('civ14.txt', 12712, 55, 0.004326620516047829)\n",
      "('q1.48.txt', 3694, 16, 0.004331348132106118)\n",
      "('curtius8.txt', 11077, 48, 0.004333303240949715)\n",
      "('pomponius3.txt', 5076, 22, 0.004334121355397951)\n",
      "('aen11.txt', 6226, 27, 0.004336652746546739)\n",
      "('psplato.sisyphus.txt', 1613, 7, 0.0043397396156230625)\n",
      "('appverg.aetna.txt', 4367, 19, 0.004350812915044653)\n",
      "('may7.txt', 2758, 12, 0.00435097897026831)\n",
      "('geo2.txt', 3673, 16, 0.004356112169888375)\n",
      "('q1.59.txt', 2294, 10, 0.004359197907585004)\n",
      "('verus.txt', 2063, 9, 0.004362578768783325)\n",
      "('historia.txt', 6647, 29, 0.004362870467880246)\n",
      "('serm4.txt', 7553, 33, 0.004369124851052562)\n",
      "('marbodus.txt', 1601, 7, 0.004372267332916927)\n",
      "('prosperus.sententiae.txt', 14634, 64, 0.004373377067104004)\n",
      "('alexsev.txt', 457, 2, 0.00437636761487965)\n",
      "('pliny.ep5.txt', 5928, 26, 0.0043859649122807015)\n",
      "('scauro.txt', 2952, 13, 0.004403794037940379)\n",
      "('civ7.txt', 11341, 50, 0.004408782294330306)\n",
      "('blas.txt', 907, 4, 0.004410143329658214)\n",
      "('conf1.txt', 5212, 23, 0.004412893323100537)\n",
      "('bede2.txt', 11723, 52, 0.004435724643862492)\n",
      "('nd3.txt', 9015, 40, 0.004437049362174155)\n",
      "('sententiae2.txt', 14195, 63, 0.004438182458612187)\n",
      "('mac.txt', 2478, 11, 0.00443906376109766)\n",
      "('may6.txt', 2249, 10, 0.004446420631391729)\n",
      "('fam14.txt', 3365, 15, 0.004457652303120356)\n",
      "('burnett.txt', 224, 1, 0.004464285714285714)\n",
      "('iordanes2.txt', 18579, 83, 0.004467409440766457)\n",
      "('annalesregnifrancorum.txt', 21711, 97, 0.004467781309013864)\n",
      "('henry2.txt', 890, 4, 0.0044943820224719105)\n",
      "('paris.txt', 667, 3, 0.004497751124437781)\n",
      "('mapps2.txt', 1776, 8, 0.0045045045045045045)\n",
      "('apuleius.deosocratis.txt', 4659, 21, 0.004507405022537025)\n",
      "('tacitus.txt', 3106, 14, 0.004507405022537025)\n",
      "('tertullian.marcionem5.txt', 18177, 82, 0.004511195466798702)\n",
      "('nd1.txt', 11292, 51, 0.004516471838469713)\n",
      "('tibullus1.txt', 5312, 24, 0.004518072289156626)\n",
      "('columella.rr6.txt', 9516, 43, 0.00451870533837747)\n",
      "('conf4.txt', 5089, 23, 0.0045195519748477105)\n",
      "('gregorytours10.txt', 14350, 65, 0.004529616724738676)\n",
      "('malaterra2.txt', 11024, 50, 0.0045355587808417995)\n",
      "('fran.txt', 660, 3, 0.004545454545454545)\n",
      "('civ19.txt', 14503, 66, 0.00455078259670413)\n",
      "('suet.persius.txt', 438, 2, 0.0045662100456621)\n",
      "('abbofloracensis.txt', 4596, 21, 0.0045691906005221935)\n",
      "('pliny.nh2.txt', 17503, 80, 0.00457064503228018)\n",
      "('seneca.suasoriae.txt', 10063, 46, 0.004571201430984796)\n",
      "('bedecontinuatio.txt', 437, 2, 0.004576659038901602)\n",
      "('hist8.txt', 7638, 35, 0.0045823514008902855)\n",
      "('fin5.txt', 11112, 51, 0.00458963282937365)\n",
      "('conf13.txt', 8693, 40, 0.004601403428045554)\n",
      "('holberg.txt', 28651, 132, 0.004607169034239643)\n",
      "('40.txt', 217, 1, 0.004608294930875576)\n",
      "('liv.per55.txt', 217, 1, 0.004608294930875576)\n",
      "('geo3.txt', 3891, 18, 0.004626060138781804)\n",
      "('gestafrancorum10.txt', 6252, 29, 0.0046385156749840055)\n",
      "('quintilian.decl.mai1.txt', 3448, 16, 0.004640371229698376)\n",
      "('bacon.hist6.txt', 6890, 32, 0.0046444121915820025)\n",
      "('esther.txt', 4089, 19, 0.0046466128637808754)\n",
      "('exodus.txt', 20627, 96, 0.004654094148446211)\n",
      "('helio.txt', 5799, 27, 0.0046559751681324365)\n",
      "('psplato.eryxias.txt', 5154, 24, 0.004656577415599534)\n",
      "('geo1.txt', 3429, 16, 0.004666083406240887)\n",
      "('praefatio.txt', 214, 1, 0.004672897196261682)\n",
      "('columella.rr9.txt', 7045, 33, 0.004684173172462739)\n",
      "('richerus3.txt', 11947, 56, 0.004687369214028627)\n",
      "('mir5.txt', 213, 1, 0.004694835680751174)\n",
      "('pliny.ep4.txt', 6162, 29, 0.004706264199935086)\n",
      "('q1.87.txt', 2544, 12, 0.0047169811320754715)\n",
      "('manasses.txt', 212, 1, 0.0047169811320754715)\n",
      "('vitruvius6.txt', 5082, 24, 0.004722550177095631)\n",
      "('claud.txt', 2963, 14, 0.004724940938238272)\n",
      "('syrus.txt', 8238, 39, 0.0047341587764020395)\n",
      "('pliny.ep7.txt', 6106, 29, 0.004749426793318048)\n",
      "('civ2.txt', 10096, 48, 0.004754358161648178)\n",
      "('serm1.txt', 7142, 34, 0.004760571268552226)\n",
      "('withof7.txt', 4201, 20, 0.00476077124494168)\n",
      "('civ21.txt', 15692, 75, 0.004779505480499617)\n",
      "('erchempert.txt', 12335, 59, 0.004783137413862991)\n",
      "('leg2.txt', 7523, 36, 0.004785325003323142)\n",
      "('q1.66.txt', 3550, 17, 0.004788732394366197)\n",
      "('andrea.txt', 2922, 14, 0.004791238877481177)\n",
      "('barometri.txt', 416, 2, 0.004807692307692308)\n",
      "('peter2.txt', 1036, 5, 0.004826254826254826)\n",
      "('tusc3.txt', 8702, 42, 0.004826476672029419)\n",
      "('suet.galba.txt', 2900, 14, 0.004827586206896552)\n",
      "('marcellinus1.txt', 9528, 46, 0.004827875734676742)\n",
      "('john.txt', 14079, 68, 0.004829888486398181)\n",
      "('maxbal.txt', 3104, 15, 0.004832474226804124)\n",
      "('suet.gram.txt', 2897, 14, 0.004832585433206766)\n",
      "('paradoxa.txt', 4343, 21, 0.004835367257655998)\n",
      "('q1.72.txt', 826, 4, 0.004842615012106538)\n",
      "('liv.per46.txt', 206, 1, 0.0048543689320388345)\n",
      "('reposianus.txt', 1236, 6, 0.0048543689320388345)\n",
      "('perp.txt', 3497, 17, 0.004861309694023448)\n",
      "('valmax3.txt', 9253, 45, 0.004863287582405706)\n",
      "('campion.plot2.txt', 3691, 18, 0.004876727174207532)\n",
      "('liv.per39.txt', 205, 1, 0.004878048780487805)\n",
      "('silius9.txt', 4296, 21, 0.004888268156424581)\n",
      "('bacon.hist9.txt', 5109, 25, 0.004893325504012527)\n",
      "('hyginus3.txt', 3678, 18, 0.004893964110929853)\n",
      "('prud.psycho.txt', 6128, 30, 0.00489556135770235)\n",
      "('gregdecretals4.txt', 29364, 144, 0.004903964037597058)\n",
      "('kempis3.txt', 20360, 100, 0.004911591355599214)\n",
      "('quintilian.institutio6.txt', 11977, 59, 0.0049261083743842365)\n",
      "('iordanes1.txt', 20292, 100, 0.004928050463236744)\n",
      "('ovid.ponto2.txt', 5061, 25, 0.0049397352301916615)\n",
      "('silius8.txt', 4441, 22, 0.004953839225399685)\n",
      "('henry3.txt', 1211, 6, 0.00495458298926507)\n",
      "('conf6.txt', 5446, 27, 0.004957767168564084)\n",
      "('phaedr4.txt', 2819, 14, 0.004966300106420717)\n",
      "('phaedr5.txt', 1004, 5, 0.0049800796812749)\n",
      "('luke.txt', 18015, 90, 0.004995836802664446)\n",
      "('val.txt', 999, 5, 0.005005005005005005)\n",
      "('epistola.txt', 998, 5, 0.00501002004008016)\n",
      "('songofsongs.txt', 1795, 9, 0.005013927576601671)\n",
      "('cato.agri.txt', 15148, 76, 0.005017163982043834)\n",
      "('johannes.txt', 12345, 62, 0.005022276225192386)\n",
      "('ruaeus.txt', 2380, 12, 0.005042016806722689)\n",
      "('alanus1.txt', 16824, 85, 0.005052306229196386)\n",
      "('oresmius.txt', 9693, 49, 0.0050551944702362525)\n",
      "('quintilian.institutio8.txt', 13252, 67, 0.005055840627829762)\n",
      "('tertullian.praescrip.txt', 8282, 42, 0.005071238831200193)\n",
      "('nepos.mil.txt', 1376, 7, 0.005087209302325582)\n",
      "('conf5.txt', 4711, 24, 0.005094459774994693)\n",
      "('sall.frag.txt', 10178, 52, 0.005109058754175673)\n",
      "('waltarius3.txt', 2543, 13, 0.005112072355485647)\n",
      "('civ13.txt', 10113, 52, 0.0051418965687728664)\n",
      "('fam9.txt', 7964, 41, 0.005148166750376695)\n",
      "('orosius1.txt', 8351, 43, 0.0051490839420428695)\n",
      "('hannes.txt', 194, 1, 0.005154639175257732)\n",
      "('liv.per80.txt', 194, 1, 0.005154639175257732)\n",
      "('histrom16.txt', 1940, 10, 0.005154639175257732)\n",
      "('tertullian.apol.txt', 19959, 103, 0.005160579187334035)\n",
      "('mensoria.txt', 578, 3, 0.005190311418685121)\n",
      "('kempis1.txt', 8430, 44, 0.005219454329774615)\n",
      "('seneca.ep16.txt', 4590, 24, 0.00522875816993464)\n",
      "('rudens.txt', 11276, 59, 0.005232351897836112)\n",
      "('divinst1.txt', 17541, 92, 0.00524485491135055)\n",
      "('gellius8.txt', 381, 2, 0.005249343832020997)\n",
      "('lucan3.txt', 4952, 26, 0.005250403877221325)\n",
      "('serm17.txt', 1714, 9, 0.005250875145857643)\n",
      "('q1.62.txt', 4755, 25, 0.005257623554153523)\n",
      "('hugo1.txt', 4178, 22, 0.005265677357587362)\n",
      "('wmconchesphil.txt', 5684, 30, 0.005277973258268825)\n",
      "('claudian.cons6.txt', 4346, 23, 0.00529222273354809)\n",
      "('bede3.txt', 16020, 85, 0.005305867665418227)\n",
      "('vitruvius10.txt', 8282, 44, 0.005312726394590679)\n",
      "('liv.per35.txt', 188, 1, 0.005319148936170213)\n",
      "('rusticus.txt', 3195, 17, 0.005320813771517997)\n",
      "('tibullus3.txt', 4504, 24, 0.0053285968028419185)\n",
      "('ovid.her18.txt', 1501, 8, 0.005329780146568954)\n",
      "('ovid.tristia1.txt', 4877, 26, 0.005331146196432233)\n",
      "('ter.adel.txt', 8242, 44, 0.005338510070371269)\n",
      "('repub5.txt', 373, 2, 0.005361930294906166)\n",
      "('silius17.txt', 4286, 23, 0.005366308912739151)\n",
      "('quintilian.institutio1.txt', 17136, 92, 0.005368814192343604)\n",
      "('bacon.sermones.txt', 42223, 227, 0.0053762167539019015)\n",
      "('matthew.txt', 16360, 88, 0.0053789731051344745)\n",
      "('venantius.txt', 742, 4, 0.005390835579514825)\n",
      "('dialogus.txt', 27600, 149, 0.005398550724637681)\n",
      "('gregdecretals2.txt', 32771, 177, 0.005401116841109517)\n",
      "('tunger.txt', 12939, 70, 0.005410000772857253)\n",
      "('bebel.txt', 18850, 102, 0.005411140583554377)\n",
      "('dares1.txt', 7930, 43, 0.005422446406052964)\n",
      "('liv.per30.txt', 184, 1, 0.005434782608695652)\n",
      "('falcandus.txt', 35693, 194, 0.0054352394026839996)\n",
      "('anast.txt', 551, 3, 0.0054446460980036296)\n",
      "('grattius.txt', 3661, 20, 0.005462988254575253)\n",
      "('gregorytours8.txt', 10614, 58, 0.00546448087431694)\n",
      "('sen.apoc.txt', 2928, 16, 0.00546448087431694)\n",
      "('sulpiciusseveruschron2.txt', 11883, 65, 0.005469999158461668)\n",
      "('vell1.txt', 3645, 20, 0.0054869684499314125)\n",
      "('gestafrancorum4.txt', 1275, 7, 0.005490196078431373)\n",
      "('lucan7.txt', 5826, 32, 0.0054926192928252664)\n",
      "('lucan8.txt', 5826, 32, 0.0054926192928252664)\n",
      "('liv.per116.txt', 182, 1, 0.005494505494505495)\n",
      "('liv.per77.txt', 182, 1, 0.005494505494505495)\n",
      "('silius5.txt', 4364, 24, 0.005499541704857928)\n",
      "('petronius1.txt', 30901, 170, 0.005501440082845216)\n",
      "('phaedr1.txt', 2178, 12, 0.005509641873278237)\n",
      "('columella.rr12.txt', 15239, 84, 0.005512172714745062)\n",
      "('sedulius4.txt', 1990, 11, 0.005527638190954774)\n",
      "('epist2.txt', 3255, 18, 0.005529953917050691)\n",
      "('q1.46.txt', 3431, 19, 0.005537744097930632)\n",
      "('epist1.txt', 6860, 38, 0.0055393586005830905)\n",
      "('varro.ll10.txt', 5054, 28, 0.00554016620498615)\n",
      "('ovid.tristia2.txt', 3787, 21, 0.005545286506469501)\n",
      "('james.txt', 1619, 9, 0.005558987029030266)\n",
      "('solinus4.txt', 7722, 43, 0.005568505568505569)\n",
      "('bacon.hist1.txt', 4844, 27, 0.005573905862923204)\n",
      "('balde2.txt', 179, 1, 0.00558659217877095)\n",
      "('solinus4a.txt', 8411, 47, 0.005587920580192605)\n",
      "('columba2.txt', 5188, 29, 0.00558982266769468)\n",
      "('seneca.ep6.txt', 6619, 37, 0.005589968273153044)\n",
      "('q1.2.txt', 2323, 13, 0.005596211795092553)\n",
      "('consulatu.txt', 534, 3, 0.0056179775280898875)\n",
      "('may3.txt', 2490, 14, 0.005622489959839358)\n",
      "('tertullian.carne.txt', 9573, 54, 0.005640864932623002)\n",
      "('hist7.txt', 14333, 81, 0.005651294216144562)\n",
      "('praelium.txt', 1060, 6, 0.005660377358490566)\n",
      "('spinoza.ethica4.txt', 16602, 94, 0.005661968437537646)\n",
      "('tusc1.txt', 12521, 71, 0.005670473604344701)\n",
      "('fam16.txt', 4583, 26, 0.005673139864717434)\n",
      "('trina.txt', 705, 4, 0.005673758865248227)\n",
      "('apuleius.dog2.txt', 5463, 31, 0.005674537799743731)\n",
      "('apuleius.florida.txt', 7925, 45, 0.0056782334384858045)\n",
      "('acad.txt', 4910, 28, 0.005702647657841141)\n",
      "('q1.5.txt', 3330, 19, 0.005705705705705706)\n",
      "('solinus5.txt', 32737, 187, 0.005712191098756758)\n",
      "('leo2.txt', 5590, 32, 0.005724508050089445)\n",
      "('tertullian.genesis.txt', 1048, 6, 0.0057251908396946565)\n",
      "('iulianus2.txt', 10649, 61, 0.0057282373931824586)\n",
      "('q1.8.txt', 2617, 15, 0.00573175391669851)\n",
      "('sententiae3.txt', 19332, 111, 0.0057417752948479206)\n",
      "('q1.43.txt', 4354, 25, 0.005741846577859439)\n",
      "('prologus.txt', 1217, 7, 0.005751848808545604)\n",
      "('smarius.txt', 1041, 6, 0.005763688760806916)\n",
      "('caeciliusbalbus.txt', 6068, 35, 0.005767963085036256)\n",
      "('bacon.distributio.txt', 3279, 19, 0.005794449527294907)\n",
      "('tertullian.marcionem4.txt', 33133, 192, 0.005794826909727462)\n",
      "('36.txt', 1035, 6, 0.005797101449275362)\n",
      "('q1.33.txt', 2759, 16, 0.005799202609641174)\n",
      "('victor.origio.txt', 4481, 26, 0.005802276277616604)\n",
      "('tertullian.testimonia.txt', 2239, 13, 0.005806163465832961)\n",
      "('annalesvedastini.txt', 7749, 45, 0.005807200929152149)\n",
      "('q1.52.txt', 1205, 7, 0.005809128630705394)\n",
      "('avianus.txt', 4130, 24, 0.005811138014527845)\n",
      "('aen4.txt', 4816, 28, 0.005813953488372093)\n",
      "('gregorytours7.txt', 11004, 64, 0.0058160668847691745)\n",
      "('tertullian.cultu2.txt', 3256, 19, 0.005835380835380835)\n",
      "('histrom15.txt', 2227, 13, 0.0058374494836102376)\n",
      "('sedulius3.txt', 2217, 13, 0.005863779882724403)\n",
      "('liv.per48.txt', 511, 3, 0.005870841487279843)\n",
      "('nepos.alc.txt', 2041, 12, 0.005879470847623714)\n",
      "('liv.per59.txt', 339, 2, 0.0058997050147492625)\n",
      "('histrom1.txt', 2880, 17, 0.005902777777777778)\n",
      "('q1.74.txt', 2709, 16, 0.005906238464377999)\n",
      "('gestafrancorum8.txt', 1184, 7, 0.0059121621621621625)\n",
      "('nepos.epam.txt', 1691, 10, 0.0059136605558840925)\n",
      "('ovid.ponto3.txt', 4899, 29, 0.005919575423555827)\n",
      "('gestafrancorum2.txt', 1515, 9, 0.005940594059405941)\n",
      "('zonaras.txt', 10268, 61, 0.0059407869107908065)\n",
      "('ovid.her7.txt', 1346, 8, 0.005943536404160475)\n",
      "('leo3.txt', 6052, 36, 0.005948446794448116)\n",
      "('digest28.txt', 25886, 154, 0.005949161709031909)\n",
      "('tertullian.pudicitia.txt', 13431, 80, 0.005956369592733229)\n",
      "('gregorytours6.txt', 13078, 78, 0.005964214711729622)\n",
      "('sedulius2.txt', 2012, 12, 0.005964214711729622)\n",
      "('vitruvius1.txt', 6036, 36, 0.005964214711729622)\n",
      "('gregdecretals1.txt', 105692, 631, 0.0059701774968777205)\n",
      "('vegetius4.txt', 5523, 33, 0.005975013579576317)\n",
      "('ovid.met14.txt', 5855, 35, 0.005977796754910333)\n",
      "('tertullian.uxor1.txt', 2007, 12, 0.005979073243647235)\n",
      "('serm19.txt', 2174, 13, 0.005979760809567618)\n",
      "('celtis.oratio.txt', 3837, 23, 0.005994266353922335)\n",
      "('silius1.txt', 4500, 27, 0.006)\n",
      "('gregdecretals5.txt', 86082, 517, 0.0060059013498757)\n",
      "('aen7.txt', 5493, 33, 0.006007646095030038)\n",
      "('wmconchesdogma.txt', 13312, 80, 0.006009615384615385)\n",
      "('conf12.txt', 7309, 44, 0.0060199753728280205)\n",
      "('macabees1.txt', 16441, 99, 0.006021531537011131)\n",
      "('varro.rr2.txt', 10769, 65, 0.00603584362522054)\n",
      "('ec7.txt', 497, 3, 0.006036217303822937)\n",
      "('amphitruo.txt', 9595, 58, 0.006044815007816571)\n",
      "('terraiam.txt', 165, 1, 0.006060606060606061)\n",
      "('serm77.txt', 2800, 17, 0.006071428571428571)\n",
      "('gregorytours5.txt', 16800, 102, 0.006071428571428571)\n",
      "('bernardcluny2.txt', 7222, 44, 0.006092495153697037)\n",
      "('serm76.txt', 1641, 10, 0.006093845216331505)\n",
      "('ephesians.txt', 2133, 13, 0.006094702297233943)\n",
      "('liv.per104.txt', 164, 1, 0.006097560975609756)\n",
      "('regula.txt', 5564, 34, 0.006110711718188354)\n",
      "('civ12.txt', 10637, 65, 0.006110745510952336)\n",
      "('gregdecretals3.txt', 91239, 558, 0.006115805740966034)\n",
      "('bacon.liber2.txt', 39387, 241, 0.0061187701525884175)\n",
      "('vir.txt', 980, 6, 0.006122448979591836)\n",
      "('conf2.txt', 2775, 17, 0.006126126126126126)\n",
      "('omnegenus.txt', 163, 1, 0.006134969325153374)\n",
      "('ovid.tristia5.txt', 5052, 31, 0.00613618368962787)\n",
      "('thom.txt', 1466, 9, 0.006139154160982265)\n",
      "('ovid.fasti1.txt', 4705, 29, 0.0061636556854410204)\n",
      "('ec9.txt', 486, 3, 0.006172839506172839)\n",
      "('sigebert.vitabrevior.txt', 1457, 9, 0.0061770761839396015)\n",
      "('dares.txt', 7761, 48, 0.006184770003865481)\n",
      "('catechizandis.txt', 14707, 91, 0.006187529747739172)\n",
      "('kempis4.txt', 7909, 49, 0.006195473511189784)\n",
      "('des.med1.txt', 1291, 8, 0.006196746707978312)\n",
      "('godfrey.epigrammata.txt', 6938, 43, 0.006197751513404439)\n",
      "('tertullian.anima.txt', 23859, 148, 0.006203109937549771)\n",
      "('sidonius9.txt', 6607, 41, 0.006205539579234145)\n",
      "('colossians.txt', 1446, 9, 0.006224066390041493)\n",
      "('tertullian.nationes1.txt', 7710, 48, 0.0062256809338521405)\n",
      "('q1.86.txt', 2087, 13, 0.006229036895064686)\n",
      "('att3.txt', 5939, 37, 0.006230005051355447)\n",
      "('prud14.txt', 642, 4, 0.006230529595015576)\n",
      "('bede5.txt', 17161, 107, 0.006235067886486802)\n",
      "('ovid.amor3.txt', 5612, 35, 0.006236635780470421)\n",
      "('manilius2.txt', 6569, 41, 0.006241437052823869)\n",
      "('tertullian.marcionem3.txt', 10555, 66, 0.006252960682141166)\n",
      "('macarius1.txt', 4793, 30, 0.006259127894846651)\n",
      "('leo1.txt', 6224, 39, 0.006266066838046273)\n",
      "('persius.txt', 4623, 29, 0.00627298291152931)\n",
      "('solinus2a.txt', 6529, 41, 0.006279675294838413)\n",
      "('ovid.tristia3.txt', 5249, 33, 0.006286911792722423)\n",
      "('manilius4.txt', 6521, 41, 0.006287379236313449)\n",
      "('liv.per56.txt', 159, 1, 0.006289308176100629)\n",
      "('aen12.txt', 6514, 41, 0.006294135707706479)\n",
      "('civ11.txt', 12064, 76, 0.0062997347480106105)\n",
      "('lucan6.txt', 5396, 34, 0.006300963676797628)\n",
      "('campion.umbra.txt', 2692, 17, 0.006315007429420505)\n",
      "('luc.txt', 949, 6, 0.006322444678609062)\n",
      "('stichus.txt', 6166, 39, 0.006325008108984755)\n",
      "('timothy1.txt', 1577, 10, 0.0063411540900443885)\n",
      "('williamapulia.txt', 17473, 111, 0.006352658387226006)\n",
      "('q1.17.txt', 2358, 15, 0.006361323155216285)\n",
      "('q1.38.txt', 1096, 7, 0.006386861313868613)\n",
      "('rutilius.txt', 4070, 26, 0.0063882063882063885)\n",
      "('deuteronomy.txt', 18781, 120, 0.006389436132261328)\n",
      "('lucan9.txt', 7348, 47, 0.0063962983124659775)\n",
      "('q1.68.txt', 2963, 19, 0.00641241984475194)\n",
      "('strat4.txt', 4513, 29, 0.006425880788832263)\n",
      "('quintilian.institutio2.txt', 14310, 92, 0.006429070580013976)\n",
      "('tibullus2.txt', 2798, 18, 0.006433166547533953)\n",
      "('sen.phoen.txt', 4186, 27, 0.006450071667462972)\n",
      "('liv.per34.txt', 155, 1, 0.0064516129032258064)\n",
      "('epistvaria.txt', 28669, 185, 0.006452963130907949)\n",
      "('cistellaria.txt', 5112, 33, 0.006455399061032864)\n",
      "('campion.plot1.txt', 4953, 32, 0.006460730870179689)\n",
      "('vitruvius7.txt', 5721, 37, 0.0064674008040552355)\n",
      "('silius13.txt', 5872, 38, 0.006471389645776567)\n",
      "('versus.txt', 309, 2, 0.006472491909385114)\n",
      "('novatian.txt', 20377, 132, 0.00647789174068803)\n",
      "('lucan10.txt', 3687, 24, 0.006509357200976403)\n",
      "('ovid.met9.txt', 5521, 36, 0.006520557869951096)\n",
      "('volovirum.txt', 153, 1, 0.006535947712418301)\n",
      "('fabulae.txt', 611, 4, 0.006546644844517185)\n",
      "('hebrews.txt', 4582, 30, 0.006547359231776516)\n",
      "('arspoet.txt', 3206, 21, 0.006550218340611353)\n",
      "('q1.71.txt', 763, 5, 0.00655307994757536)\n",
      "('xanten.txt', 6713, 44, 0.006554446596156711)\n",
      "('walton.txt', 3502, 23, 0.0065676756139348945)\n",
      "('suet.nero.txt', 8066, 53, 0.0065707909744607)\n",
      "('apicius4.txt', 2583, 17, 0.006581494386372435)\n",
      "('tertullian.ieiunio.txt', 5924, 39, 0.006583389601620527)\n",
      "('marcellinus2.txt', 10023, 66, 0.006584854833882071)\n",
      "('jeremiah.txt', 29568, 195, 0.006594967532467532)\n",
      "('hist11.txt', 6971, 46, 0.006598766317601492)\n",
      "('nepos.thras.txt', 606, 4, 0.006600660066006601)\n",
      "('digest45.txt', 19235, 127, 0.006602547439563296)\n",
      "('silius14.txt', 4390, 29, 0.006605922551252847)\n",
      "('petrarch.rom.txt', 1967, 13, 0.0066090493136756485)\n",
      "('hugo6.txt', 5139, 34, 0.0066160731659856)\n",
      "('quum.txt', 302, 2, 0.006622516556291391)\n",
      "('conf3.txt', 3920, 26, 0.0066326530612244895)\n",
      "('valeriusflaccus6.txt', 5270, 35, 0.006641366223908918)\n",
      "('galatians.txt', 2107, 14, 0.006644518272425249)\n",
      "('tertullian.deiudicio.txt', 2556, 17, 0.006651017214397496)\n",
      "('landor1810.txt', 451, 3, 0.0066518847006651885)\n",
      "('tertullian.spect.txt', 6308, 42, 0.006658211794546607)\n",
      "('axio.txt', 2402, 16, 0.006661115736885929)\n",
      "('sedulius5.txt', 2852, 19, 0.006661991584852735)\n",
      "('epidicus.txt', 6153, 41, 0.006663416219730213)\n",
      "('liv.per61.txt', 150, 1, 0.006666666666666667)\n",
      "('repellenda.txt', 1197, 8, 0.006683375104427736)\n",
      "('nepos.kings.txt', 447, 3, 0.006711409395973154)\n",
      "('tertullian.baptismo.txt', 4317, 29, 0.006717627982395182)\n",
      "('petrarch.ep1.txt', 893, 6, 0.006718924972004479)\n",
      "('ovid.met1.txt', 5347, 36, 0.00673274733495418)\n",
      "('silius2.txt', 4598, 31, 0.006742061765985211)\n",
      "('prosperus.epistola.txt', 2219, 15, 0.0067598017124831005)\n",
      "('peter1.txt', 1626, 11, 0.006765067650676507)\n",
      "('ovid.amor2.txt', 5312, 36, 0.00677710843373494)\n",
      "('amos.txt', 2799, 19, 0.006788138620936048)\n",
      "('ovid.fasti2.txt', 5596, 38, 0.006790564689063617)\n",
      "('innocent1.txt', 14283, 97, 0.006791290345165581)\n",
      "('aen5.txt', 5886, 40, 0.0067957866123003734)\n",
      "('vitruvius9.txt', 6033, 41, 0.006795955577656224)\n",
      "('corinthians2.txt', 4263, 29, 0.006802721088435374)\n",
      "('ovid.fasti6.txt', 5279, 36, 0.006819473385110817)\n",
      "('q1.70.txt', 2639, 18, 0.006820765441455097)\n",
      "('propertius1.txt', 4396, 30, 0.006824385805277525)\n",
      "('liv.per5.txt', 293, 2, 0.006825938566552901)\n",
      "('bacon.hist11.txt', 3074, 21, 0.0068314899154196486)\n",
      "('tertullian.iudaeos.txt', 11118, 76, 0.006835761827666847)\n",
      "('carm2.txt', 2486, 17, 0.006838294448913918)\n",
      "('berengar.txt', 5240, 36, 0.006870229007633588)\n",
      "('trin3.txt', 7130, 49, 0.006872370266479663)\n",
      "('oratio.stephani.txt', 871, 6, 0.006888633754305396)\n",
      "('luther.praef.txt', 2467, 17, 0.006890960680989056)\n",
      "('sphaer.txt', 435, 3, 0.006896551724137931)\n",
      "('carus.txt', 2746, 19, 0.006919155134741442)\n",
      "('tertullian.nationes2.txt', 7081, 49, 0.006919926564044627)\n",
      "('varro.ll5.txt', 10816, 75, 0.006934171597633136)\n",
      "('varro.rr3.txt', 10084, 70, 0.006941689805632686)\n",
      "('tertullian.pallio.txt', 3457, 24, 0.0069424356378362745)\n",
      "('silius11.txt', 4032, 28, 0.006944444444444444)\n",
      "('q1.67.txt', 2444, 17, 0.006955810147299509)\n",
      "('nepos.pel.txt', 718, 5, 0.006963788300835654)\n",
      "('paulo.txt', 430, 3, 0.0069767441860465115)\n",
      "('obadiah.txt', 429, 3, 0.006993006993006993)\n",
      "('mir9.txt', 572, 4, 0.006993006993006993)\n",
      "('digest30.txt', 15703, 110, 0.007005030885817997)\n",
      "('samuel1.txt', 18415, 129, 0.007005158837903883)\n",
      "('serm78.txt', 1142, 8, 0.0070052539404553416)\n",
      "('claudian.proserp3.txt', 2855, 20, 0.0070052539404553416)\n",
      "('ovid.met15.txt', 5989, 42, 0.007012856904324595)\n",
      "('gregorytours9.txt', 14500, 102, 0.00703448275862069)\n",
      "('sigebert.script.txt', 8100, 57, 0.007037037037037037)\n",
      "('debury.txt', 14917, 105, 0.0070389488503050214)\n",
      "('philip.txt', 1562, 11, 0.007042253521126761)\n",
      "('liv.per41.txt', 142, 1, 0.007042253521126761)\n",
      "('claudian.ruf1.txt', 2555, 18, 0.0070450097847358124)\n",
      "('ovid.tristia4.txt', 4534, 32, 0.0070577856197618)\n",
      "('ovid.her17.txt', 1840, 13, 0.007065217391304348)\n",
      "('thomasedessa.txt', 10748, 76, 0.0070710829921845925)\n",
      "('baldo.txt', 9158, 65, 0.0070976195675911775)\n",
      "('q1.10.txt', 3519, 25, 0.007104290991759022)\n",
      "('vegius.txt', 4355, 31, 0.00711825487944891)\n",
      "('carmeninvictoriam.txt', 1826, 13, 0.007119386637458927)\n",
      "('alexio.txt', 1262, 9, 0.0071315372424722665)\n",
      "('ovid.ponto4.txt', 6161, 44, 0.007141697776335011)\n",
      "('liv.per103.txt', 140, 1, 0.007142857142857143)\n",
      "('lotichius.txt', 140, 1, 0.007142857142857143)\n",
      "('maidstone.txt', 1677, 12, 0.007155635062611807)\n",
      "('columella.rr10.txt', 3072, 22, 0.007161458333333333)\n",
      "('paulinus.poemata.txt', 24990, 179, 0.007162865146058423)\n",
      "('calpurniussiculus.txt', 5165, 37, 0.007163601161665053)\n",
      "('petrarch.numa.txt', 837, 6, 0.007168458781362007)\n",
      "('ep.txt', 3067, 22, 0.007173133355070101)\n",
      "('appvergcomp.txt', 16432, 118, 0.007181110029211295)\n",
      "('24.txt', 1809, 13, 0.007186290768380321)\n",
      "('nic.txt', 1249, 9, 0.007205764611689352)\n",
      "('gwinne2.txt', 3883, 28, 0.0072109193922225085)\n",
      "('q1.16.txt', 4158, 30, 0.007215007215007215)\n",
      "('sen.medea.txt', 5811, 42, 0.007227671657201859)\n",
      "('bacchides.txt', 9673, 70, 0.00723663806471622)\n",
      "('firmus.txt', 2347, 17, 0.007243289305496378)\n",
      "('judith.txt', 6626, 48, 0.00724418955629339)\n",
      "('octavia.txt', 5383, 39, 0.0072450306520527585)\n",
      "('nepos.dion.txt', 1518, 11, 0.007246376811594203)\n",
      "('tertullian.patientia.txt', 4551, 33, 0.007251153592617007)\n",
      "('digest7.txt', 15530, 113, 0.007276239536381198)\n",
      "('q1.39.txt', 6734, 49, 0.007276507276507277)\n",
      "('daniel.txt', 10848, 79, 0.007282448377581121)\n",
      "('quadragesima2.txt', 960, 7, 0.007291666666666667)\n",
      "('serm2.txt', 7402, 54, 0.007295325587679005)\n",
      "('silvae5.txt', 5755, 42, 0.007298001737619461)\n",
      "('nepos.lys.txt', 548, 4, 0.0072992700729927005)\n",
      "('mart3.txt', 3971, 29, 0.007302946361118106)\n",
      "('solinus3.txt', 8747, 64, 0.007316794329484394)\n",
      "('gregorytours4.txt', 12695, 93, 0.007325718786923986)\n",
      "('mercator.txt', 8322, 61, 0.0073299687575102135)\n",
      "('geo4.txt', 3816, 28, 0.007337526205450734)\n",
      "('silvae4.txt', 4497, 33, 0.00733822548365577)\n",
      "('q1.45.txt', 4901, 36, 0.007345439706182412)\n",
      "('bernardcluny1.txt', 9252, 68, 0.007349762213575443)\n",
      "('corinthians1.txt', 6386, 47, 0.007359849671155653)\n",
      "('carm4.txt', 2988, 22, 0.007362784471218206)\n",
      "('commodianus2.txt', 7599, 56, 0.007369390709303856)\n",
      "('digest22.txt', 9224, 68, 0.007372072853425846)\n",
      "('claudian.olyb.txt', 1763, 13, 0.00737379466817924)\n",
      "('bacon.liber1.txt', 19116, 141, 0.0073760200878844944)\n",
      "('prop2.txt', 8800, 65, 0.007386363636363636)\n",
      "('tertullian.scorpiace.txt', 7979, 59, 0.00739441032710866)\n",
      "('tertullian.martyres.txt', 1486, 11, 0.007402422611036339)\n",
      "('mart8.txt', 4178, 31, 0.007419818094782192)\n",
      "('hugo3.txt', 4445, 33, 0.007424071991001125)\n",
      "('strat1.txt', 6059, 45, 0.007426968146558838)\n",
      "('bacon.hist5.txt', 4173, 31, 0.007428708363287803)\n",
      "('truculentus.txt', 8067, 60, 0.007437709185570844)\n",
      "('gestafrancorum3.txt', 800, 6, 0.0075)\n",
      "('janus1.txt', 2400, 18, 0.0075)\n",
      "('junillus.txt', 10266, 77, 0.007500487044613286)\n",
      "('waltarius2.txt', 4132, 31, 0.007502420135527589)\n",
      "('may4.txt', 2932, 22, 0.007503410641200546)\n",
      "('gwinne5.1.txt', 3331, 25, 0.007505253677574302)\n",
      "('columba1.txt', 9722, 73, 0.00750874305698416)\n",
      "('pomponius1.txt', 5856, 44, 0.007513661202185792)\n",
      "('liv.per107.txt', 133, 1, 0.007518796992481203)\n",
      "('gwinne3.txt', 5981, 45, 0.007523825447249624)\n",
      "('silius12.txt', 4913, 37, 0.007531040097699979)\n",
      "('nemesianus4.txt', 531, 4, 0.007532956685499058)\n",
      "('haggai.txt', 796, 6, 0.007537688442211055)\n",
      "('q1.55.txt', 1857, 14, 0.007539041464728056)\n",
      "('aen10.txt', 6192, 47, 0.007590439276485788)\n",
      "('mostellaria.txt', 9087, 69, 0.0075932651039947174)\n",
      "('aen9.txt', 5527, 42, 0.007599059164103492)\n",
      "('bede4.txt', 17196, 131, 0.007618050709467318)\n",
      "('sententiae1.txt', 12202, 93, 0.007621701360432716)\n",
      "('valeriusflaccus7.txt', 4708, 36, 0.0076465590484282074)\n",
      "('pervig.txt', 652, 5, 0.007668711656441718)\n",
      "('mysteriis.txt', 4824, 37, 0.007669983416252073)\n",
      "('conflictus.txt', 391, 3, 0.0076726342710997444)\n",
      "('kempis2.txt', 5075, 39, 0.007684729064039409)\n",
      "('blesensis.txt', 7157, 55, 0.0076847841274276935)\n",
      "('asinaria.txt', 7806, 60, 0.007686395080707148)\n",
      "('ovid.her3.txt', 1040, 8, 0.007692307692307693)\n",
      "('civ18.txt', 13760, 106, 0.007703488372093024)\n",
      "('gwinne4.txt', 5711, 44, 0.007704430047277185)\n",
      "('archpoet.txt', 4279, 33, 0.007712082262210797)\n",
      "('mon1.txt', 5310, 41, 0.007721280602636535)\n",
      "('ovid.fasti3.txt', 5693, 44, 0.007728789741788161)\n",
      "('manilius1.txt', 6209, 48, 0.007730713480431631)\n",
      "('liv.per113.txt', 129, 1, 0.007751937984496124)\n",
      "('mir8.txt', 387, 3, 0.007751937984496124)\n",
      "('campion.epigr1.txt', 6939, 54, 0.007782101167315175)\n",
      "('ovid.met3.txt', 5121, 40, 0.007810974419058777)\n",
      "('biggs.txt', 6273, 49, 0.007811254583134066)\n",
      "('timothy2.txt', 1152, 9, 0.0078125)\n",
      "('creeds.txt', 895, 7, 0.00782122905027933)\n",
      "('carm3.txt', 4475, 35, 0.00782122905027933)\n",
      "('ovid.fasti5.txt', 4730, 37, 0.007822410147991543)\n",
      "('innocent2.txt', 3834, 30, 0.00782472613458529)\n",
      "('romans.txt', 6502, 51, 0.007843740387573054)\n",
      "('varro.ll8.txt', 4967, 39, 0.007851822025367425)\n",
      "('hyginus1.txt', 2036, 16, 0.007858546168958742)\n",
      "('prop4.txt', 6098, 48, 0.00787143325680551)\n",
      "('digest46.txt', 25103, 198, 0.007887503485639167)\n",
      "('q1.69.txt', 1774, 14, 0.007891770011273957)\n",
      "('civ17.txt', 14061, 111, 0.007894175378707062)\n",
      "('vitruvius3.txt', 4047, 32, 0.007907091672844082)\n",
      "('hist10.txt', 8336, 66, 0.00791746641074856)\n",
      "('lucan5.txt', 5428, 43, 0.007921886514369933)\n",
      "('silius7.txt', 4923, 39, 0.007921998781230956)\n",
      "('q1.83.txt', 2271, 18, 0.007926023778071334)\n",
      "('quintilian.institutio9.txt', 19177, 152, 0.007926161547687334)\n",
      "('hydatiuschronicon.txt', 5172, 41, 0.007927300850734726)\n",
      "('mart6.txt', 3784, 30, 0.007928118393234672)\n",
      "('laude.txt', 1765, 14, 0.00793201133144476)\n",
      "('ovid.amor1.txt', 5039, 40, 0.007938082952966859)\n",
      "('q1.58.txt', 3527, 28, 0.00793875815140346)\n",
      "('tertullian.herm.txt', 11199, 89, 0.00794713813733369)\n",
      "('appverg.ciris.txt', 3522, 28, 0.007950028392958546)\n",
      "('hyginus4.txt', 5393, 43, 0.007973298720563694)\n",
      "('columbus.txt', 2255, 18, 0.007982261640798226)\n",
      "('q1.9.txt', 1377, 11, 0.007988380537400145)\n",
      "('aen8.txt', 4882, 39, 0.007988529291274069)\n",
      "('tertullian.corona.txt', 4880, 39, 0.007991803278688524)\n",
      "('solinus3a.txt', 8106, 65, 0.008018751542067604)\n",
      "('ant.txt', 1239, 10, 0.008071025020177562)\n",
      "('mart11.txt', 4953, 40, 0.00807591358772461)\n",
      "('may1.txt', 2228, 18, 0.00807899461400359)\n",
      "('ovid.ibis.txt', 4207, 34, 0.008081768481102924)\n",
      "('valmax9.txt', 9021, 73, 0.00809222924287773)\n",
      "('tertullian.oratione.txt', 4448, 36, 0.008093525179856115)\n",
      "('ovid.rem.txt', 5312, 43, 0.00809487951807229)\n",
      "('digest48.txt', 31247, 253, 0.008096777290619899)\n",
      "('gwinne1.txt', 5680, 46, 0.008098591549295774)\n",
      "('des.ep.txt', 1110, 9, 0.008108108108108109)\n",
      "('septsap.txt', 8013, 65, 0.008111818295270187)\n",
      "('don.txt', 4059, 33, 0.008130081300813009)\n",
      "('aulularia.txt', 6765, 55, 0.008130081300813009)\n",
      "('ovid.artis3.txt', 5287, 43, 0.008133156799697372)\n",
      "('germanicus.txt', 4793, 39, 0.008136866263300646)\n",
      "('mart14.txt', 3192, 26, 0.008145363408521303)\n",
      "('landor.1858.txt', 2946, 24, 0.008146639511201629)\n",
      "('walter3.txt', 1836, 15, 0.008169934640522876)\n",
      "('ovid.her2.txt', 979, 8, 0.008171603677221655)\n",
      "('tobia.txt', 5008, 41, 0.008186900958466454)\n",
      "('att12.txt', 9279, 76, 0.008190537773466969)\n",
      "('prosperus.rufinum.txt', 4029, 33, 0.008190618019359643)\n",
      "('des.med2.txt', 2311, 19, 0.008221549112938122)\n",
      "('decretum.txt', 1946, 16, 0.008221993833504625)\n",
      "('iacob.txt', 485, 4, 0.008247422680412371)\n",
      "('sen.thyestes.txt', 6302, 52, 0.008251348778165662)\n",
      "('revelation.txt', 8362, 69, 0.008251614446304711)\n",
      "('ovid.her11.txt', 847, 7, 0.008264462809917356)\n",
      "('spinoza.ethica5.txt', 7501, 62, 0.008265564591387815)\n",
      "('ovid.met2.txt', 6042, 50, 0.00827540549486925)\n",
      "('ovid.her21.txt', 1691, 14, 0.00827912477823773)\n",
      "('ovid.met6.txt', 4946, 41, 0.008289526890416498)\n",
      "('silius3.txt', 4581, 38, 0.008295132067234229)\n",
      "('ovid.met10.txt', 5180, 43, 0.008301158301158301)\n",
      "('ovid.met12.txt', 4335, 36, 0.008304498269896194)\n",
      "('ovid.artis1.txt', 5056, 42, 0.008306962025316455)\n",
      "('solinus2.txt', 9022, 75, 0.008313012635779207)\n",
      "('vitapauli.txt', 2404, 20, 0.008319467554076539)\n",
      "('cella.txt', 240, 2, 0.008333333333333333)\n",
      "('owen.txt', 480, 4, 0.008333333333333333)\n",
      "('pascoli.laur.txt', 840, 7, 0.008333333333333333)\n",
      "('civ16.txt', 17636, 147, 0.00833522340666818)\n",
      "('pliny.nhpr.txt', 1559, 13, 0.008338678640153944)\n",
      "('baruch.txt', 3596, 30, 0.008342602892102336)\n",
      "('q1.42.txt', 3715, 31, 0.008344549125168237)\n",
      "('aen1.txt', 5140, 43, 0.008365758754863813)\n",
      "('foedusaeternum.txt', 478, 4, 0.008368200836820083)\n",
      "('vicentius.txt', 16602, 139, 0.008372485242741838)\n",
      "('q1.34.txt', 2504, 21, 0.008386581469648562)\n",
      "('may5.txt', 1907, 16, 0.008390141583639224)\n",
      "('ovid.her9.txt', 1072, 9, 0.008395522388059701)\n",
      "('valeriusflaccus8.txt', 3330, 28, 0.008408408408408409)\n",
      "('sen.phaedra.txt', 7372, 62, 0.008410200759631037)\n",
      "('commodianus1.txt', 7235, 61, 0.008431237042156186)\n",
      "('cotta.txt', 1895, 16, 0.008443271767810026)\n",
      "('ampelius.txt', 7224, 61, 0.008444075304540421)\n",
      "('serm6.txt', 1299, 11, 0.008468052347959968)\n",
      "('kalila.txt', 29966, 254, 0.008476273109524127)\n",
      "('valeriusflaccus3.txt', 5189, 44, 0.008479475814222393)\n",
      "('hist2.txt', 4363, 37, 0.008480403392161356)\n",
      "('kings1.txt', 17562, 149, 0.008484227308962533)\n",
      "('gwinne5.3.txt', 2121, 18, 0.008486562942008486)\n",
      "('superbia.txt', 1296, 11, 0.008487654320987654)\n",
      "('mon2.txt', 6478, 55, 0.008490274776165483)\n",
      "('corvinus1.txt', 353, 3, 0.0084985835694051)\n",
      "('avienus.periegesis.txt', 9036, 77, 0.008521469676848163)\n",
      "('mart13.txt', 1877, 16, 0.008524240809802876)\n",
      "('suet.claudius.txt', 6674, 57, 0.008540605334132454)\n",
      "('nivis.txt', 234, 2, 0.008547008547008548)\n",
      "('ovid.met4.txt', 5499, 47, 0.008547008547008548)\n",
      "('prologi.txt', 2922, 25, 0.008555783709787816)\n",
      "('montanus.txt', 818, 7, 0.008557457212713936)\n",
      "('mart1.txt', 5140, 44, 0.008560311284046693)\n",
      "('ovid.met7.txt', 5943, 51, 0.008581524482584554)\n",
      "('captivi.txt', 8496, 73, 0.008592278719397363)\n",
      "('mir2.txt', 349, 3, 0.008595988538681949)\n",
      "('histbrit.txt', 9749, 84, 0.008616268335213869)\n",
      "('epistaustras.txt', 4176, 36, 0.008620689655172414)\n",
      "('concilium2.txt', 1740, 15, 0.008620689655172414)\n",
      "('prataiam.txt', 116, 1, 0.008620689655172414)\n",
      "('prud8.txt', 116, 1, 0.008620689655172414)\n",
      "('withof3.txt', 116, 1, 0.008620689655172414)\n",
      "('navagero.txt', 6958, 60, 0.00862316757688991)\n",
      "('digest33.txt', 20906, 181, 0.008657801588060845)\n",
      "('fletcher.txt', 5888, 51, 0.008661684782608696)\n",
      "('pascoli.iug.txt', 922, 8, 0.008676789587852495)\n",
      "('sedulius1.txt', 2418, 21, 0.008684863523573201)\n",
      "('persa.txt', 7366, 64, 0.008688569101276133)\n",
      "('suet.crispus.txt', 115, 1, 0.008695652173913044)\n",
      "('poree.txt', 7453, 65, 0.008721320273715282)\n",
      "('raoul.txt', 35989, 314, 0.008724888160271194)\n",
      "('manilius3.txt', 4685, 41, 0.008751334044823906)\n",
      "('egeria1.txt', 9368, 82, 0.008753202391118702)\n",
      "('catullus.txt', 13123, 115, 0.00876324011277909)\n",
      "('q1.50.txt', 3421, 30, 0.008769365682548963)\n",
      "('silvae2.txt', 5213, 46, 0.008824093612123537)\n",
      "('balbus.txt', 2148, 19, 0.008845437616387336)\n",
      "('apicius5.txt', 1356, 12, 0.008849557522123894)\n",
      "('mart4.txt', 3950, 35, 0.008860759493670886)\n",
      "('ovid.fasti4.txt', 6207, 55, 0.008860963428387304)\n",
      "('exhortatio.txt', 1467, 13, 0.008861622358554875)\n",
      "('asserius.txt', 12181, 108, 0.008866267137345046)\n",
      "('q1.19.txt', 7203, 64, 0.008885186727752325)\n",
      "('cas.txt', 7310, 65, 0.008891928864569083)\n",
      "('vitruvius4.txt', 4048, 36, 0.008893280632411068)\n",
      "('strat3.txt', 4156, 37, 0.00890279114533205)\n",
      "('lim.txt', 560, 5, 0.008928571428571428)\n",
      "('att5.txt', 7390, 66, 0.008930987821380243)\n",
      "('nepos.timo.txt', 671, 6, 0.00894187779433681)\n",
      "('quintilian.institutio3.txt', 14175, 127, 0.008959435626102293)\n",
      "('pascoli.catull.txt', 2006, 18, 0.00897308075772682)\n",
      "('vida.txt', 4457, 40, 0.008974646623289208)\n",
      "('machinae.txt', 557, 5, 0.008976660682226212)\n",
      "('des.syn.txt', 891, 8, 0.008978675645342313)\n",
      "('prud3.txt', 891, 8, 0.008978675645342313)\n",
      "('ec8.txt', 779, 7, 0.008985879332477536)\n",
      "('ovid.met5.txt', 4664, 42, 0.009005145797598629)\n",
      "('suet.lucan.txt', 222, 2, 0.009009009009009009)\n",
      "('gregory7.txt', 16088, 145, 0.009012928891098955)\n",
      "('27.txt', 664, 6, 0.009036144578313253)\n",
      "('suet.dom.txt', 3646, 33, 0.009051014810751509)\n",
      "('nepos.phoc.txt', 552, 5, 0.009057971014492754)\n",
      "('hist1.txt', 5934, 54, 0.00910010111223458)\n",
      "('ovid.met8.txt', 5933, 54, 0.009101634923310299)\n",
      "('mart7.txt', 4491, 41, 0.009129369850812736)\n",
      "('withof6.txt', 3286, 30, 0.009129640900791236)\n",
      "('ovid.her13.txt', 1094, 10, 0.009140767824497258)\n",
      "('silvae3.txt', 5357, 49, 0.009146910584282247)\n",
      "('petrarchmedicus.txt', 22409, 205, 0.009148110134321031)\n",
      "('kings2.txt', 16273, 149, 0.009156271123947643)\n",
      "('macarius.txt', 1417, 13, 0.009174311926605505)\n",
      "('fratrem3.txt', 5336, 49, 0.009182908545727137)\n",
      "('menaechmi.txt', 9142, 84, 0.009188361408882083)\n",
      "('q1.56.txt', 1849, 17, 0.009194159004867496)\n",
      "('att4.txt', 6740, 62, 0.009198813056379823)\n",
      "('campion.elegies.txt', 4021, 37, 0.00920169112161154)\n",
      "('pax.txt', 1195, 11, 0.009205020920502092)\n",
      "('suet.horace.txt', 543, 5, 0.009208103130755065)\n",
      "('nepos.con.txt', 760, 7, 0.009210526315789473)\n",
      "('piccolomini.ep6.txt', 542, 5, 0.00922509225092251)\n",
      "('fide.txt', 5742, 53, 0.009230233368164403)\n",
      "('prud6.txt', 758, 7, 0.009234828496042216)\n",
      "('seneca.fragmenta.txt', 324, 3, 0.009259259259259259)\n",
      "('digest18.txt', 28364, 263, 0.009272317021576647)\n",
      "('dumdiane.txt', 215, 2, 0.009302325581395349)\n",
      "('notitia1.txt', 5372, 50, 0.009307520476545048)\n",
      "('q1.25.txt', 4074, 38, 0.009327442317133039)\n",
      "('piccolomini.ep5.txt', 4930, 46, 0.009330628803245436)\n",
      "('gelliuscapitula.txt', 6644, 62, 0.009331727874774232)\n",
      "('bonaventura.itinerarium.txt', 9963, 93, 0.009334537789822343)\n",
      "('liv.per42.txt', 107, 1, 0.009345794392523364)\n",
      "('hymns.txt', 962, 9, 0.009355509355509356)\n",
      "('aen2.txt', 5449, 51, 0.009359515507432557)\n",
      "('thessalonians2.txt', 743, 7, 0.009421265141318977)\n",
      "('poenulus.txt', 10991, 104, 0.009462287325993995)\n",
      "('theb11.txt', 5275, 50, 0.009478672985781991)\n",
      "('bedeproverbs.txt', 8526, 81, 0.009500351864883884)\n",
      "('ec10.txt', 526, 5, 0.009505703422053232)\n",
      "('contraioannem.txt', 13037, 124, 0.009511390657359823)\n",
      "('tertullian.cultu1.txt', 1891, 18, 0.009518773135906928)\n",
      "('institutes.proem.txt', 420, 4, 0.009523809523809525)\n",
      "('waardenburg.txt', 525, 5, 0.009523809523809525)\n",
      "('q1.1.txt', 4618, 44, 0.009527934170636639)\n",
      "('ovid.her15.txt', 1469, 14, 0.009530292716133424)\n",
      "('valeriusflaccus4.txt', 5349, 51, 0.009534492428491307)\n",
      "('digest35.txt', 22691, 217, 0.009563262967696443)\n",
      "('digest34.txt', 20485, 196, 0.00956797656822065)\n",
      "('suet.rhet.txt', 1254, 12, 0.009569377990430622)\n",
      "('quadragesima1.txt', 1149, 11, 0.009573542210617928)\n",
      "('ovid.her16.txt', 2502, 24, 0.009592326139088728)\n",
      "('theb6.txt', 6359, 61, 0.00959270325522881)\n",
      "('ovid.artis2.txt', 4896, 47, 0.00959967320261438)\n",
      "('ter.hecyra.txt', 7596, 73, 0.009610321221695629)\n",
      "('pliny.nh3.txt', 8630, 83, 0.009617612977983778)\n",
      "('mart12.txt', 4450, 43, 0.009662921348314606)\n",
      "('tertullian.idololatria.txt', 6930, 67, 0.009668109668109668)\n",
      "('theb9.txt', 6191, 60, 0.009691487643353255)\n",
      "('coll.txt', 515, 5, 0.009708737864077669)\n",
      "('paulinus.ausonium.txt', 103, 1, 0.009708737864077669)\n",
      "('sannazaro1.txt', 9672, 94, 0.009718775847808106)\n",
      "('campion.epigr2.txt', 8735, 85, 0.009730967372638809)\n",
      "('hist9.txt', 11601, 113, 0.009740539608654427)\n",
      "('porphyrius.txt', 5133, 50, 0.00974089226573154)\n",
      "('q1.40.txt', 2566, 25, 0.009742790335151987)\n",
      "('att8.txt', 5337, 52, 0.009743301480232341)\n",
      "('aen6.txt', 6151, 60, 0.009754511461550968)\n",
      "('tertullian.propheta.txt', 717, 7, 0.009762900976290097)\n",
      "('pliny.ep1.txt', 6248, 61, 0.009763124199743917)\n",
      "('nepos.them.txt', 1741, 17, 0.009764503159103963)\n",
      "('pomponius2.txt', 5928, 58, 0.009784075573549258)\n",
      "('ter.eunuchus.txt', 9248, 91, 0.009839965397923875)\n",
      "('serm7.txt', 1726, 17, 0.009849362688296639)\n",
      "('chronicles2.txt', 18346, 181, 0.00986591082524801)\n",
      "('varro.ll6.txt', 6173, 61, 0.00988174307468006)\n",
      "('ovid.met13.txt', 6762, 67, 0.009908311150547175)\n",
      "('sannazaro2.txt', 807, 8, 0.009913258983890954)\n",
      "('campion.thamesin.txt', 1905, 19, 0.009973753280839895)\n",
      "('q1.27.txt', 2803, 28, 0.009989297181591153)\n",
      "('claudian.proserp2.txt', 2601, 26, 0.009996155324875048)\n",
      "('ec2.txt', 500, 5, 0.01)\n",
      "('pseudolus.txt', 10785, 108, 0.010013908205841447)\n",
      "('tertullian.gentium.txt', 699, 7, 0.010014306151645207)\n",
      "('valeriusflaccus2.txt', 4687, 47, 0.010027736291871132)\n",
      "('apicius2.txt', 796, 8, 0.010050251256281407)\n",
      "('gwinne5.4.txt', 3283, 33, 0.010051781906792567)\n",
      "('q1.47.txt', 1890, 19, 0.010052910052910053)\n",
      "('fulgentius4.txt', 1989, 20, 0.010055304172951232)\n",
      "('ovid.her14.txt', 895, 9, 0.01005586592178771)\n",
      "('p1.txt', 497, 5, 0.01006036217303823)\n",
      "('nepos.cha.txt', 497, 5, 0.01006036217303823)\n",
      "('q1.63.txt', 5355, 54, 0.010084033613445379)\n",
      "('anselmepistula.txt', 694, 7, 0.010086455331412104)\n",
      "('aus.mos.txt', 3072, 31, 0.010091145833333334)\n",
      "('apuleius.apol.txt', 22107, 224, 0.01013253720541005)\n",
      "('spinoza.ethica3.txt', 16357, 166, 0.010148560249434493)\n",
      "('q1.54.txt', 2364, 24, 0.01015228426395939)\n",
      "('ebulo.txt', 11010, 112, 0.010172570390554041)\n",
      "('mart10.txt', 5307, 54, 0.010175240248728096)\n",
      "('enn.txt', 1179, 12, 0.010178117048346057)\n",
      "('att15.txt', 6872, 70, 0.01018626309662398)\n",
      "('vitacaroli.txt', 15208, 155, 0.010192004208311414)\n",
      "('mart2.txt', 3532, 36, 0.010192525481313703)\n",
      "('spinoza.ethica1.txt', 10300, 105, 0.010194174757281554)\n",
      "('john1.txt', 1857, 19, 0.010231556273559504)\n",
      "('digest24.txt', 16595, 170, 0.010244049412473637)\n",
      "('suet.virgil.txt', 1464, 15, 0.010245901639344262)\n",
      "('landor1795.txt', 2533, 26, 0.010264508487958943)\n",
      "('strat2.txt', 8568, 88, 0.01027077497665733)\n",
      "('mirabilia.txt', 4575, 47, 0.010273224043715848)\n",
      "('carmina.txt', 973, 10, 0.010277492291880781)\n",
      "('q1.57.txt', 3208, 33, 0.010286783042394014)\n",
      "('ter.phormio.txt', 9505, 98, 0.010310362966859547)\n",
      "('ovid.her4.txt', 1163, 12, 0.010318142734307825)\n",
      "('digest49.txt', 19228, 199, 0.01034949032660703)\n",
      "('genesis.txt', 25971, 269, 0.010357706672827384)\n",
      "('forsett2.txt', 3379, 35, 0.010358094110683634)\n",
      "('theb5.txt', 5190, 54, 0.010404624277456647)\n",
      "('may2.txt', 2391, 25, 0.010455876202425763)\n",
      "('ec1.txt', 572, 6, 0.01048951048951049)\n",
      "('appvergculex.txt', 2669, 28, 0.01049082053203447)\n",
      "('sen.agamemnon.txt', 5702, 60, 0.01052262364082778)\n",
      "('q1.41.txt', 4080, 43, 0.01053921568627451)\n",
      "('prop3.txt', 6352, 67, 0.010547858942065492)\n",
      "('agnes.txt', 10040, 106, 0.01055776892430279)\n",
      "('theb2.txt', 5005, 53, 0.01058941058941059)\n",
      "('titum.txt', 661, 7, 0.01059001512859304)\n",
      "('priapea.txt', 3490, 37, 0.010601719197707736)\n",
      "('valeriusflaccus5.txt', 4896, 52, 0.010620915032679739)\n",
      "('26.txt', 659, 7, 0.010622154779969651)\n",
      "('patricius1.txt', 3294, 35, 0.010625379477838493)\n",
      "('liv.per105.txt', 94, 1, 0.010638297872340425)\n",
      "('ec3.txt', 846, 9, 0.010638297872340425)\n",
      "('sen.oedipus.txt', 6091, 65, 0.010671482515186341)\n",
      "('theophanes.txt', 281, 3, 0.010676156583629894)\n",
      "('alanus2.txt', 28605, 306, 0.010697430519140011)\n",
      "('appverg.catalepton.txt', 1302, 14, 0.010752688172043012)\n",
      "('corpuschristi.txt', 93, 1, 0.010752688172043012)\n",
      "('adso.txt', 1948, 21, 0.010780287474332649)\n",
      "('manilius5.txt', 5008, 54, 0.010782747603833865)\n",
      "('concilium1.txt', 2411, 26, 0.010783907092492741)\n",
      "('theb10.txt', 6382, 69, 0.01081165778752742)\n",
      "('psplato.halcyon.txt', 1202, 13, 0.010815307820299502)\n",
      "('silvae1.txt', 5441, 59, 0.010843594927403052)\n",
      "('suet.pliny.txt', 92, 1, 0.010869565217391304)\n",
      "('mart9.txt', 5323, 58, 0.010896111215479993)\n",
      "('digest38.txt', 26384, 288, 0.010915706488781079)\n",
      "('piccolomini.carmen.txt', 366, 4, 0.01092896174863388)\n",
      "('digest41.txt', 22231, 245, 0.011020646844496424)\n",
      "('q1.64.txt', 2896, 32, 0.011049723756906077)\n",
      "('judges.txt', 13178, 146, 0.011079071179238124)\n",
      "('estas.txt', 90, 1, 0.011111111111111112)\n",
      "('liv.per31.txt', 180, 2, 0.011111111111111112)\n",
      "('bedepraef.txt', 717, 8, 0.011157601115760111)\n",
      "('brevechronicon.txt', 1165, 13, 0.011158798283261802)\n",
      "('ovid.her8.txt', 806, 9, 0.011166253101736972)\n",
      "('digest1.txt', 15806, 177, 0.011198279134505884)\n",
      "('q1.12.txt', 7589, 85, 0.011200421662933193)\n",
      "('musavenit.txt', 89, 1, 0.011235955056179775)\n",
      "('theb4.txt', 5776, 65, 0.011253462603878116)\n",
      "('sen.hercules.txt', 11524, 130, 0.011280805275945853)\n",
      "('q1.14.txt', 9741, 110, 0.011292475105225337)\n",
      "('nepos.aris.txt', 353, 4, 0.0113314447592068)\n",
      "('q1.30.txt', 2732, 31, 0.011346998535871157)\n",
      "('q1.31.txt', 2643, 30, 0.011350737797956867)\n",
      "('att14.txt', 6939, 79, 0.011384925781812941)\n",
      "('psplato.amatores.txt', 2370, 27, 0.01139240506329114)\n",
      "('fulgentius2.txt', 4207, 48, 0.011409555502733539)\n",
      "('psplato.iusto.txt', 1314, 15, 0.01141552511415525)\n",
      "('gregorytours1.txt', 7348, 84, 0.011431682090364725)\n",
      "('varro.ll7.txt', 5331, 61, 0.011442506096417183)\n",
      "('mon3.txt', 7333, 84, 0.011455066139369972)\n",
      "('trinummus.txt', 9600, 110, 0.011458333333333333)\n",
      "('henrysettimello.txt', 6525, 75, 0.011494252873563218)\n",
      "('carm1.txt', 4072, 47, 0.011542239685658153)\n",
      "('q1.73.txt', 1557, 18, 0.011560693641618497)\n",
      "('anon.nev.txt', 6996, 81, 0.011578044596912522)\n",
      "('cato.frag.txt', 1120, 13, 0.011607142857142858)\n",
      "('planctus.txt', 429, 5, 0.011655011655011656)\n",
      "('q1.32.txt', 3172, 37, 0.011664564943253467)\n",
      "('liv.per47.txt', 171, 2, 0.011695906432748537)\n",
      "('liv.per67.txt', 171, 2, 0.011695906432748537)\n",
      "('numbers.txt', 19882, 233, 0.011719142943365858)\n",
      "('fulgentius5.txt', 4005, 47, 0.011735330836454432)\n",
      "('hugo4.txt', 4165, 49, 0.011764705882352941)\n",
      "('levis.txt', 85, 1, 0.011764705882352941)\n",
      "('rhetores.txt', 85, 1, 0.011764705882352941)\n",
      "('eucherius.txt', 3904, 46, 0.0117827868852459)\n",
      "('civ10.txt', 14339, 169, 0.011786038077969175)\n",
      "('carmsaec.txt', 339, 4, 0.011799410029498525)\n",
      "('ovid.her12.txt', 1437, 17, 0.011830201809324982)\n",
      "('civ9.txt', 7393, 88, 0.011903151629920195)\n",
      "('ferraria.txt', 2435, 29, 0.01190965092402464)\n",
      "('magnacarta.txt', 3605, 43, 0.011927877947295423)\n",
      "('ovid.her6.txt', 1085, 13, 0.011981566820276499)\n",
      "('q1.51.txt', 2085, 25, 0.011990407673860911)\n",
      "('digest36.txt', 23770, 286, 0.012031973075305007)\n",
      "('nobilis.txt', 83, 1, 0.012048192771084338)\n",
      "('iulianus1.txt', 16145, 196, 0.012139981418395788)\n",
      "('liv.per38.txt', 247, 3, 0.012145748987854251)\n",
      "('aen3.txt', 4770, 58, 0.012159329140461216)\n",
      "('garcilaso.txt', 982, 12, 0.012219959266802444)\n",
      "('digest8.txt', 12429, 152, 0.012229463351838442)\n",
      "('valeriusflaccus1.txt', 5967, 73, 0.012233953410424)\n",
      "('vico.orat6.txt', 3099, 38, 0.012262020006453695)\n",
      "('q1.44.txt', 2199, 27, 0.01227830832196453)\n",
      "('hist.txt', 9997, 123, 0.012303691107332199)\n",
      "('civ8.txt', 11599, 143, 0.012328649021467368)\n",
      "('liv.per93.txt', 81, 1, 0.012345679012345678)\n",
      "('aus.sept.sent.txt', 322, 4, 0.012422360248447204)\n",
      "('ter.andria.txt', 8599, 107, 0.012443307361321083)\n",
      "('ter.heauton.txt', 9058, 113, 0.012475160079487745)\n",
      "('eugenius.txt', 80, 1, 0.0125)\n",
      "('conf.txt', 9551, 120, 0.012564129410532929)\n",
      "('seneca.ep17-18.txt', 8906, 112, 0.012575791601167752)\n",
      "('theb1.txt', 4743, 60, 0.01265022137887413)\n",
      "('theb7.txt', 5610, 71, 0.01265597147950089)\n",
      "('esdras1.txt', 8041, 102, 0.012684989429175475)\n",
      "('mart5.txt', 3777, 48, 0.012708498808578236)\n",
      "('theb3.txt', 4949, 63, 0.01272984441301273)\n",
      "('boskovic.txt', 3057, 39, 0.012757605495583905)\n",
      "('ave.phoen.txt', 1096, 14, 0.012773722627737226)\n",
      "('ovid.met11.txt', 5474, 70, 0.01278772378516624)\n",
      "('rimbaud.txt', 391, 5, 0.01278772378516624)\n",
      "('sulpiciusseveruschron1.txt', 12865, 165, 0.012825495530509134)\n",
      "('piccolomini.ep2.txt', 1006, 13, 0.012922465208747515)\n",
      "('claudian.proserp1.txt', 1856, 24, 0.01293103448275862)\n",
      "('vulgar.txt', 5870, 76, 0.012947189097103918)\n",
      "('mir1.txt', 771, 10, 0.01297016861219196)\n",
      "('thessalonians1.txt', 1387, 18, 0.012977649603460706)\n",
      "('albertanus.sermo1.txt', 2844, 37, 0.013009845288326301)\n",
      "('att1.txt', 9451, 123, 0.01301449582054809)\n",
      "('digest40.txt', 32029, 417, 0.013019451122420306)\n",
      "('psplato.minos.txt', 2455, 32, 0.013034623217922607)\n",
      "('gellius11.txt', 4603, 60, 0.01303497718878992)\n",
      "('samuel2.txt', 14814, 194, 0.013095720264614554)\n",
      "('gioacchino.txt', 20919, 274, 0.013098140446484057)\n",
      "('carm.bur.txt', 229, 3, 0.013100436681222707)\n",
      "('angilbert.txt', 304, 4, 0.013157894736842105)\n",
      "('ipsavivere.txt', 76, 1, 0.013157894736842105)\n",
      "('mart.spec.txt', 1366, 18, 0.013177159590043924)\n",
      "('vitruvius5.txt', 6069, 80, 0.013181743285549514)\n",
      "('des.med4.txt', 2271, 30, 0.013210039630118891)\n",
      "('censorinus.txt', 10280, 136, 0.013229571984435798)\n",
      "('piccolomini.ep4.txt', 3394, 45, 0.013258691809074838)\n",
      "('des.pr.txt', 754, 10, 0.013262599469496022)\n",
      "('expositio.txt', 6404, 85, 0.013272954403497813)\n",
      "('theb12.txt', 5573, 74, 0.013278306118787008)\n",
      "('mapps1.txt', 1506, 20, 0.013280212483399735)\n",
      "('att16.txt', 6918, 92, 0.0132986412257878)\n",
      "('patricius2.txt', 3004, 40, 0.013315579227696404)\n",
      "('ovid.her5.txt', 1049, 14, 0.01334604385128694)\n",
      "('des.med5.txt', 1793, 24, 0.013385387618516454)\n",
      "('hyginus5.txt', 10265, 138, 0.013443740867023867)\n",
      "('ency.fides.txt', 27000, 363, 0.013444444444444445)\n",
      "('gellius4.txt', 5427, 73, 0.013451262207481112)\n",
      "('hyginus2.txt', 10628, 143, 0.013455024463680843)\n",
      "('galileo.sid.txt', 9722, 132, 0.013577453198930261)\n",
      "('digest12.txt', 16489, 224, 0.013584814118503244)\n",
      "('digest29.txt', 20748, 282, 0.013591671486408328)\n",
      "('att10.txt', 6472, 88, 0.013597033374536464)\n",
      "('halley.txt', 294, 4, 0.013605442176870748)\n",
      "('serm79.txt', 220, 3, 0.013636363636363636)\n",
      "('withof2.txt', 220, 3, 0.013636363636363636)\n",
      "('des.med6.txt', 4096, 56, 0.013671875)\n",
      "('12tables.txt', 658, 9, 0.013677811550151976)\n",
      "('moriae.txt', 24977, 342, 0.013692597189414261)\n",
      "('lucernarium.txt', 73, 1, 0.0136986301369863)\n",
      "('seneca.ep20.txt', 7581, 104, 0.013718506793299038)\n",
      "('q1.22.txt', 2621, 36, 0.013735215566577643)\n",
      "('john2.txt', 218, 3, 0.013761467889908258)\n",
      "('legenda.stephani.txt', 3261, 45, 0.013799448022079117)\n",
      "('withof1.txt', 217, 3, 0.013824884792626729)\n",
      "('q1.11.txt', 2230, 31, 0.01390134529147982)\n",
      "('ovid.her10.txt', 1006, 14, 0.013916500994035786)\n",
      "('serm74.txt', 1002, 14, 0.013972055888223553)\n",
      "('q1.20.txt', 2713, 38, 0.014006634721710284)\n",
      "('tertullian.haereses.txt', 2567, 36, 0.014024152707440592)\n",
      "('ec6.txt', 570, 8, 0.014035087719298246)\n",
      "('diesirae.txt', 213, 3, 0.014084507042253521)\n",
      "('att7.txt', 9082, 128, 0.014093811935696983)\n",
      "('jude.txt', 425, 6, 0.01411764705882353)\n",
      "('forsett1.txt', 4671, 66, 0.014129736673089274)\n",
      "('q1.61.txt', 1626, 23, 0.014145141451414513)\n",
      "('john3.txt', 212, 3, 0.014150943396226415)\n",
      "('digest32.txt', 15615, 221, 0.014153057957092539)\n",
      "('q1.6.txt', 1762, 25, 0.014188422247446084)\n",
      "('bill.rights.txt', 422, 6, 0.014218009478672985)\n",
      "('albertanus.sermo4.txt', 1616, 23, 0.014232673267326733)\n",
      "('digest44.txt', 14643, 209, 0.014273031482619682)\n",
      "('protospatarius.txt', 4123, 59, 0.014309968469560999)\n",
      "('ec5.txt', 625, 9, 0.0144)\n",
      "('apuleius.dog1.txt', 3541, 51, 0.014402711098559728)\n",
      "('theb8.txt', 5241, 76, 0.01450104941804999)\n",
      "('ilias.txt', 7017, 102, 0.01453612654980761)\n",
      "('digest15.txt', 11678, 170, 0.014557287206713478)\n",
      "('q1.29.txt', 3275, 48, 0.01465648854961832)\n",
      "('achilleid1.txt', 6549, 96, 0.014658726523133303)\n",
      "('walter6.txt', 613, 9, 0.01468189233278956)\n",
      "('liv.per100.txt', 68, 1, 0.014705882352941176)\n",
      "('joshua.txt', 12774, 188, 0.014717394708000627)\n",
      "('musica.txt', 1357, 20, 0.014738393515106854)\n",
      "('q1.24.txt', 1423, 21, 0.014757554462403373)\n",
      "('digest19.txt', 18085, 267, 0.014763616256566215)\n",
      "('tertullian.valentinianos.txt', 6357, 94, 0.014786849142677364)\n",
      "('piccolomini.ep3.txt', 1621, 24, 0.014805675508945095)\n",
      "('luther.95.txt', 1619, 24, 0.014823965410747375)\n",
      "('bultelius2.txt', 4316, 64, 0.014828544949026877)\n",
      "('egeria2.txt', 8210, 122, 0.014859926918392205)\n",
      "('liberpontificalis1.txt', 13914, 207, 0.014877102199223804)\n",
      "('nemesianus3.txt', 468, 7, 0.014957264957264958)\n",
      "('resurr.txt', 732, 11, 0.015027322404371584)\n",
      "('hugo2.txt', 5054, 76, 0.015037593984962405)\n",
      "('landor1806.txt', 3118, 47, 0.015073765234124438)\n",
      "('origo.txt', 995, 15, 0.01507537688442211)\n",
      "('arbroath.txt', 1059, 16, 0.015108593012275733)\n",
      "('digest31.txt', 13075, 199, 0.015219885277246653)\n",
      "('tertullian.scapulam.txt', 1510, 23, 0.015231788079470199)\n",
      "('epitaphium.txt', 196, 3, 0.015306122448979591)\n",
      "('nemesianus1.txt', 588, 9, 0.015306122448979591)\n",
      "('hydatiusfasti.txt', 5941, 91, 0.015317286652078774)\n",
      "('piccolomini.ep1.txt', 651, 10, 0.015360983102918587)\n",
      "('liv.per78.txt', 65, 1, 0.015384615384615385)\n",
      "('walter5.txt', 650, 10, 0.015384615384615385)\n",
      "('nepos.paus.txt', 904, 14, 0.015486725663716814)\n",
      "('philemon.txt', 322, 5, 0.015527950310559006)\n",
      "('commodianus3.txt', 128, 2, 0.015625)\n",
      "('walter11.txt', 320, 5, 0.015625)\n",
      "('sabinus2.txt', 698, 11, 0.015759312320916905)\n",
      "('theod04.txt', 10513, 166, 0.01578997431751165)\n",
      "('notitia2.txt', 4238, 67, 0.015809344030202926)\n",
      "('digest21.txt', 16729, 265, 0.015840755574152668)\n",
      "('rumor.txt', 126, 2, 0.015873015873015872)\n",
      "('gellius5.txt', 5768, 92, 0.0159500693481276)\n",
      "('withof5.txt', 1000, 16, 0.016)\n",
      "('digest47.txt', 28046, 449, 0.016009413107038437)\n",
      "('gellius6.txt', 7119, 114, 0.016013485040033713)\n",
      "('corvinus2.txt', 684, 11, 0.01608187134502924)\n",
      "('nithardus3.txt', 3035, 49, 0.01614497528830313)\n",
      "('sabinus3.txt', 619, 10, 0.01615508885298869)\n",
      "('digest23.txt', 16910, 274, 0.016203429923122414)\n",
      "('q1.21.txt', 2156, 35, 0.016233766233766232)\n",
      "('sedulius.solis.txt', 308, 5, 0.016233766233766232)\n",
      "('tempusest.txt', 184, 3, 0.016304347826086956)\n",
      "('codex7.txt', 35256, 576, 0.016337644656228726)\n",
      "('q1.23.txt', 5236, 86, 0.016424751718869365)\n",
      "('att13.txt', 9549, 157, 0.01644151220023039)\n",
      "('digest20.txt', 10084, 166, 0.016461721539071796)\n",
      "('theodolus.txt', 2182, 36, 0.016498625114573784)\n",
      "('orationes.txt', 1511, 25, 0.01654533421575116)\n",
      "('walter9.txt', 422, 7, 0.016587677725118485)\n",
      "('digest37.txt', 22667, 378, 0.016676225349627212)\n",
      "('mir7.txt', 179, 3, 0.01675977653631285)\n",
      "('anon.deramis.txt', 119, 2, 0.01680672268907563)\n",
      "('albertanus.sermo3.txt', 3386, 57, 0.01683402244536326)\n",
      "('fulgentius1.txt', 5158, 87, 0.016867002714230323)\n",
      "('walter7.txt', 768, 13, 0.016927083333333332)\n",
      "('vulgar2.txt', 5535, 94, 0.016982836495031618)\n",
      "('q1.3.txt', 4473, 76, 0.016990833892242343)\n",
      "('digest42.txt', 15615, 267, 0.017098943323727186)\n",
      "('codex9.txt', 22102, 378, 0.017102524658401955)\n",
      "('oratio.txt', 7402, 127, 0.01715752499324507)\n",
      "('digest5.txt', 17509, 302, 0.017248272317094065)\n",
      "('q1.18.txt', 2777, 48, 0.017284839755131436)\n",
      "('diravi.txt', 115, 2, 0.017391304347826087)\n",
      "('digest39.txt', 23873, 416, 0.017425543501026264)\n",
      "('codex10.txt', 20588, 359, 0.017437342141053042)\n",
      "('bultelius1.txt', 401, 7, 0.017456359102244388)\n",
      "('hosea.txt', 3494, 61, 0.017458500286204923)\n",
      "('digest16.txt', 18576, 325, 0.01749569336778639)\n",
      "('digest17.txt', 16047, 281, 0.01751106125755593)\n",
      "('albertanus.liberconsol.txt', 19212, 337, 0.017541120133250054)\n",
      "('digest25.txt', 6943, 122, 0.017571654904220076)\n",
      "('alfonsi.disciplina.txt', 14817, 262, 0.017682391847202536)\n",
      "('cinna.txt', 113, 2, 0.017699115044247787)\n",
      "('liv.per86.txt', 113, 2, 0.017699115044247787)\n",
      "('albertanus2.txt', 15467, 274, 0.017715135449667033)\n",
      "('theod14.txt', 8598, 153, 0.01779483600837404)\n",
      "('theod05.txt', 7243, 130, 0.01794836393759492)\n",
      "('milton.quintnov.txt', 1492, 27, 0.018096514745308313)\n",
      "('mir4.txt', 330, 6, 0.01818181818181818)\n",
      "('digest13.txt', 11480, 209, 0.018205574912891986)\n",
      "('walter1.txt', 710, 13, 0.018309859154929577)\n",
      "('pliny.nh4.txt', 7413, 136, 0.01834614865776339)\n",
      "('ruth.txt', 1827, 34, 0.018609742747673783)\n",
      "('vestiunt.txt', 107, 2, 0.018691588785046728)\n",
      "('nehemiah.txt', 7434, 139, 0.01869787463007802)\n",
      "('frodebertus.txt', 960, 18, 0.01875)\n",
      "('gellius3.txt', 6052, 114, 0.018836748182419035)\n",
      "('codex6.txt', 48605, 919, 0.018907519802489454)\n",
      "('codex4.txt', 36728, 697, 0.018977346983228054)\n",
      "('q1.65.txt', 2732, 52, 0.01903367496339678)\n",
      "('albertanus.sermo2.txt', 3183, 61, 0.01916431039899466)\n",
      "('codex11.txt', 20910, 402, 0.019225251076040173)\n",
      "('colman.txt', 260, 5, 0.019230769230769232)\n",
      "('codex12.txt', 27642, 532, 0.019246074813689312)\n",
      "('digest9.txt', 13621, 263, 0.019308420820791424)\n",
      "('theod10.txt', 11599, 224, 0.01931200965600483)\n",
      "('theod03.txt', 10084, 195, 0.019337564458548194)\n",
      "('newton.scholium.txt', 1340, 26, 0.019402985074626865)\n",
      "('inquisitio.txt', 4483, 87, 0.0194066473343743)\n",
      "('liv.per99.txt', 103, 2, 0.019417475728155338)\n",
      "('nemesianus2.txt', 611, 12, 0.019639934533551555)\n",
      "('vidularia.txt', 763, 15, 0.019659239842726082)\n",
      "('sabinus1.txt', 863, 17, 0.019698725376593278)\n",
      "('campion.misc.txt', 910, 18, 0.01978021978021978)\n",
      "('naevius.txt', 755, 15, 0.019867549668874173)\n",
      "('theod01.txt', 11172, 222, 0.01987110633727175)\n",
      "('godfrey.epigrammatahist.txt', 1406, 28, 0.01991465149359886)\n",
      "('achilleid2.txt', 1143, 23, 0.020122484689413824)\n",
      "('walter10.txt', 397, 8, 0.020151133501259445)\n",
      "('theod09.txt', 22547, 455, 0.020180068301769636)\n",
      "('arnulf.txt', 198, 4, 0.020202020202020204)\n",
      "('digest10.txt', 13353, 270, 0.02022017524151876)\n",
      "('varro.frag.txt', 1632, 33, 0.02022058823529412)\n",
      "('epitaphs.txt', 1530, 31, 0.020261437908496733)\n",
      "('curculio.txt', 6336, 129, 0.020359848484848484)\n",
      "('theod02.txt', 12917, 263, 0.020360764883486877)\n",
      "('digest3.txt', 15142, 309, 0.020406815480121517)\n",
      "('fulbert.txt', 98, 2, 0.02040816326530612)\n",
      "('columella.arbor.txt', 5854, 121, 0.020669627605056373)\n",
      "('newton.regulae.txt', 529, 11, 0.020793950850661626)\n",
      "('des.med3.txt', 4133, 86, 0.020808129687878055)\n",
      "('digest4.txt', 25018, 521, 0.02082500599568311)\n",
      "('albertanus1.txt', 5204, 109, 0.02094542659492698)\n",
      "('testamentum.txt', 285, 6, 0.021052631578947368)\n",
      "('codex3.txt', 24629, 520, 0.02111332169393804)\n",
      "('digest2.txt', 18244, 386, 0.021157640868230652)\n",
      "('walter12.txt', 472, 10, 0.0211864406779661)\n",
      "('q1.4.txt', 1792, 38, 0.021205357142857144)\n",
      "('claud.inscr.txt', 704, 15, 0.02130681818181818)\n",
      "('fronto.txt', 37139, 792, 0.021325291472576)\n",
      "('pliny.nh5.txt', 8809, 188, 0.02134180951299807)\n",
      "('walter4.txt', 701, 15, 0.021398002853067047)\n",
      "('codex5.txt', 42821, 925, 0.02160155064104061)\n",
      "('scottus.txt', 228, 5, 0.021929824561403508)\n",
      "('digest50.txt', 30978, 682, 0.022015623991219577)\n",
      "('theod08.txt', 18605, 411, 0.02209083579682881)\n",
      "('marullo.txt', 362, 8, 0.022099447513812154)\n",
      "('att2.txt', 8895, 199, 0.02237211916807195)\n",
      "('albertanus4.txt', 7182, 161, 0.022417153996101363)\n",
      "('albertanus.arsloquendi.txt', 5698, 128, 0.022464022464022463)\n",
      "('hrabanus.txt', 178, 4, 0.02247191011235955)\n",
      "('bacon.titlepage.txt', 44, 1, 0.022727272727272728)\n",
      "('walter2.txt', 615, 14, 0.022764227642276424)\n",
      "('theod13.txt', 9741, 222, 0.022790267939636587)\n",
      "('digest6.txt', 7236, 165, 0.022802653399668325)\n",
      "('garland.txt', 3199, 73, 0.022819631134729602)\n",
      "('theod07.txt', 14133, 325, 0.022995825373239936)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('q1.13.txt', 8686, 200, 0.023025558369790467)\n",
      "('q1.28.txt', 2763, 64, 0.02316322837495476)\n",
      "('digest14.txt', 10011, 232, 0.02317450804115473)\n",
      "('theod11.txt', 25347, 589, 0.02323746399968438)\n",
      "('theod16.txt', 19590, 457, 0.023328228688106176)\n",
      "('letabundus.txt', 128, 3, 0.0234375)\n",
      "('codex2.txt', 24579, 582, 0.023678750152569268)\n",
      "('spinoza.ethica2.txt', 12650, 300, 0.023715415019762844)\n",
      "('albertanus.sermo.txt', 2509, 60, 0.023913909924272617)\n",
      "('albertanus3.txt', 9946, 240, 0.024130303639654134)\n",
      "('walter8.txt', 455, 11, 0.024175824175824177)\n",
      "('apicius3.txt', 1074, 26, 0.024208566108007448)\n",
      "('liv.per81.txt', 41, 1, 0.024390243902439025)\n",
      "('digest11.txt', 9633, 235, 0.024395307796117514)\n",
      "('codex8.txt', 31513, 774, 0.024561292165138197)\n",
      "('q1.26.txt', 1099, 27, 0.02456778889899909)\n",
      "('apuleius.mundo.txt', 6643, 165, 0.02483817552310703)\n",
      "('suet.tibullus.txt', 80, 2, 0.025)\n",
      "('1644.txt', 3235, 81, 0.025038639876352395)\n",
      "('theod15.txt', 8260, 210, 0.025423728813559324)\n",
      "('theod06.txt', 17630, 454, 0.025751559841179807)\n",
      "('theod12.txt', 20577, 535, 0.025999902804101666)\n",
      "('ficino.txt', 3144, 82, 0.02608142493638677)\n",
      "('att9.txt', 9077, 237, 0.02610994822077779)\n",
      "('ec4.txt', 421, 11, 0.026128266033254157)\n",
      "('potatores.txt', 152, 4, 0.02631578947368421)\n",
      "('scaliger.txt', 38, 1, 0.02631578947368421)\n",
      "('apicius1.txt', 1095, 29, 0.026484018264840183)\n",
      "('gellius13.txt', 8906, 236, 0.026498989445317762)\n",
      "('digest43.txt', 56350, 1499, 0.026601597160603372)\n",
      "('ezra.txt', 5158, 139, 0.026948429623885228)\n",
      "('codex1.txt', 43289, 1173, 0.027096953036568183)\n",
      "('pliny.ep9.txt', 6340, 173, 0.02728706624605678)\n",
      "('pastourelles.txt', 291, 8, 0.027491408934707903)\n",
      "('gauss.txt', 8465, 234, 0.027643236857649144)\n",
      "('epitomecononiana.txt', 5008, 139, 0.0277555910543131)\n",
      "('liv.per6.txt', 107, 3, 0.028037383177570093)\n",
      "('liv.per139.txt', 35, 1, 0.02857142857142857)\n",
      "('gellius20.txt', 4014, 115, 0.028649725959143)\n",
      "('ovid.her1.txt', 762, 22, 0.028871391076115485)\n",
      "('catalogueliberien.txt', 922, 27, 0.02928416485900217)\n",
      "('lhomond.historiae.txt', 5281, 155, 0.029350501798901724)\n",
      "('digest26.txt', 19208, 590, 0.03071636817992503)\n",
      "('comes.txt', 65, 2, 0.03076923076923077)\n",
      "('gellius9.txt', 5030, 156, 0.031013916500994037)\n",
      "('iamdulcis.txt', 193, 6, 0.031088082901554404)\n",
      "('chronicles1.txt', 14789, 460, 0.03110419906687403)\n",
      "('gellius7.txt', 4309, 137, 0.03179391970294732)\n",
      "('fulgentius3.txt', 3859, 123, 0.031873542368489244)\n",
      "('epitomefeliciana.txt', 5831, 186, 0.031898473675184356)\n",
      "('liv.per109.txt', 94, 3, 0.031914893617021274)\n",
      "('luscinia.txt', 90, 3, 0.03333333333333333)\n",
      "('poems.txt', 240, 8, 0.03333333333333333)\n",
      "('hymni.txt', 1564, 53, 0.03388746803069054)\n",
      "('q1.15.txt', 1794, 61, 0.03400222965440357)\n",
      "('miles.txt', 13064, 450, 0.03444580526638089)\n",
      "('att6.txt', 6704, 239, 0.03565035799522673)\n",
      "('gellius2.txt', 9639, 344, 0.03568834941383961)\n",
      "('fabe.txt', 4348, 160, 0.03679852805887764)\n",
      "('rec.txt', 104, 4, 0.038461538461538464)\n",
      "('dicchristi.txt', 77, 3, 0.03896103896103896)\n",
      "('petroniusfrag.txt', 1115, 44, 0.039461883408071746)\n",
      "('withof4.txt', 76, 3, 0.039473684210526314)\n",
      "('newton.capita.txt', 252, 10, 0.03968253968253968)\n",
      "('gellius1.txt', 9996, 403, 0.040316126450580234)\n",
      "('withof.txt', 49, 2, 0.04081632653061224)\n",
      "('hebet.txt', 94, 4, 0.0425531914893617)\n",
      "('pliny.nh1.txt', 13293, 581, 0.04370721432332807)\n",
      "('newton.leges.txt', 88, 4, 0.045454545454545456)\n",
      "('gelliuspraef.txt', 1050, 50, 0.047619047619047616)\n",
      "('obitu.txt', 483, 23, 0.047619047619047616)\n",
      "('inscriptions.txt', 669, 32, 0.04783258594917788)\n",
      "('mir3.txt', 233, 12, 0.05150214592274678)\n",
      "('gellius10.txt', 7238, 373, 0.05153357281016856)\n",
      "('sequentia.txt', 146, 8, 0.0547945205479452)\n",
      "('resgestae1.txt', 2996, 181, 0.06041388518024032)\n",
      "('suscipeflos.txt', 66, 4, 0.06060606060606061)\n",
      "('scbaccanalibus.txt', 661, 41, 0.06202723146747353)\n",
      "('sigebert.virgin.txt', 73, 5, 0.0684931506849315)\n",
      "('seneca.ep22.txt', 194, 14, 0.07216494845360824)\n",
      "('indices.txt', 152, 12, 0.07894736842105263)\n",
      "('iabervocius.txt', 614, 56, 0.09120521172638436)\n",
      "('index.txt', 84, 8, 0.09523809523809523)\n",
      "('sicmeafata.txt', 124, 13, 0.10483870967741936)\n",
      "('carmenarvale.txt', 85, 10, 0.11764705882352941)\n",
      "('digest27.txt', 18156, 2156, 0.11874862304472351)\n",
      "('andronicus.txt', 282, 37, 0.13120567375886524)\n",
      "('andecavis.txt', 143, 20, 0.13986013986013987)\n",
      "('simedignetur.txt', 42, 7, 0.16666666666666666)\n",
      "('carmensaliare.txt', 42, 8, 0.19047619047619047)\n"
     ]
    }
   ],
   "source": [
    "print(len(corpus_files ))\n",
    "rankings = [(key, val[0], val[1], val[1]/val[0]) for key, val in corpus_files.items()]\n",
    "rankings.sort(key=lambda x: x[3])\n",
    "for rank in rankings:\n",
    "    print (rank)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# That's all for now folks!"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}