QuestionClassifier

contributors (sort by surname)

The university of manchester Postgards - March 2022

Coursework 1 for text mining comp61332

Xinyi Ouyang - Chuhan Qiu - Mingchen Wan - Zhangli Wang - Mochuan Zhan

pipeline

word tokenization -> word embedding -> sentence representation(BOW, BiLSTM) -> training classifier(NN)

folder structure

.
├── README.md
├── data
│   ├── dev.txt
│   ├── glove.small.txt
│   ├── raw_data.txt
│   ├── stopword.txt
│   ├── train.txt
│   ├── test.txt
│   └── vocabulary.txt
├── document
│   ├── README.md
│   ├── document.md
│   └── document.pdf
├── src
│   ├── utility
│   │   ├── __init__.py
│   │   ├── file_loader.py
│   │   └── pre_train.py
│   ├── __init__.py
│   ├── biLSTM.py
│   ├── bow.py
│   ├── config.ini
│   ├── model.py
│   └── question_classifier.py  
└──

environment

System environment for testing Manchester computer science virtual machine Ubuntu (64.bit) CSImage 2122 v15 PGT 4GB RAM CPU 3

hardware environment Huawei Matebook14 2019 windows 11, with 8-gen Core i5 CPU and 8GB RAM

Training dataset 5500-labeled questions

Testing dataset TREC 10

run