Made the query params that support better scoring default

adsabs · Aug 28, 2019 · 8667bf6 · 8667bf6
1 parent d8e3e05
commit 8667bf6
Show file tree

Hide file tree

Showing 6 changed files with 41 additions and 12 deletions.
diff --git a/...c/java/org/apache/lucene/queryparser/flexible/aqp/builders/AqpAdsabsSubQueryProvider.java b/...c/java/org/apache/lucene/queryparser/flexible/aqp/builders/AqpAdsabsSubQueryProvider.java
@@ -764,12 +764,15 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 
 				LuceneCacheWrapper<NumericDocValues> boostWrapper = getLuceneCache(fp, "cite_read_boost");
 
-				SecondOrderQuery outerQuery = new SecondOrderQuery( // references
+				SecondOrderQuery outerQuery = 
+				 new SecondOrderQuery( // references
 						new SecondOrderQuery( // topn
-								new SecondOrderQuery(innerQuery, // classic_relevance
-										new SecondOrderCollectorAdsClassicScoringFormula(referencesWrapper, boostWrapper)), 
+								//new SecondOrderQuery(innerQuery, // classic_relevance
+								//		new SecondOrderCollectorAdsClassicScoringFormula(referencesWrapper, boostWrapper)),
+								    innerQuery,
 										new SecondOrderCollectorTopN(200)),
 										new SecondOrderCollectorCitesRAM(referencesWrapper));
+
 				outerQuery.getcollector().setFinalValueType(FinalValueType.ABS_COUNT_NORM);
 				return outerQuery;
 			};
@@ -841,12 +844,15 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 
 				LuceneCacheWrapper<NumericDocValues> boostWrapper = getLuceneCache(fp, "cite_read_boost");
 
-				SecondOrderQuery outerQuery = new SecondOrderQuery( // citations
+				SecondOrderQuery outerQuery = 
+				 new SecondOrderQuery( // citations
 						new SecondOrderQuery( // topn
-								new SecondOrderQuery(innerQuery, // classic_relevance
-										new SecondOrderCollectorAdsClassicScoringFormula(citationsWrapper, boostWrapper)), 
+						    innerQuery,
+								//new SecondOrderQuery(innerQuery, // classic_relevance
+								//		new SecondOrderCollectorAdsClassicScoringFormula(citationsWrapper, boostWrapper)), 
 										new SecondOrderCollectorTopN(200)),
 										new SecondOrderCollectorCitedBy(citationsWrapper));
+
 				outerQuery.getcollector().setFinalValueType(FinalValueType.ABS_COUNT);
 				return outerQuery;
 			};

diff --git a/...he/lucene/queryparser/flexible/aqp/processors/AqpAdsabsFieldNodePreAnalysisProcessor.java b/...he/lucene/queryparser/flexible/aqp/processors/AqpAdsabsFieldNodePreAnalysisProcessor.java
@@ -116,7 +116,7 @@ protected QueryNode postProcessNode(QueryNode node) throws QueryNodeException {
       }
 
       Map<String, Float> statFields = getStaticFields();
-      if (statFields.containsKey(field)) {
+      if (statFields.containsKey(field) && node.getParent() != null && !(node.getParent() instanceof TermRangeQueryNode)) {
         node = new AqpConstantQueryNode(node);
         Float boost = statFields.get(field);
         if (boost != null && boost != 1.0f) {

diff --git a/contrib/adsabs/src/test/org/adsabs/TestAdsAllFields.java b/contrib/adsabs/src/test/org/adsabs/TestAdsAllFields.java
@@ -410,19 +410,22 @@ public void test() throws Exception {
 		 */
 		assertQ(req("q", "author:\"Einstein, A\"", 
 		    "aqp.constant_scoring", "author^13 title^12",
+		    "aqp.classic_scoring.modifier", "0.48",
 		    "fl", "recid,score"),
 				"//*[@numFound='1']",
 				"//doc/int[@name='recid'][.='100']",
-				"//doc/float[@name='score'][.='13.0']"
+				"//doc/float[@name='score'][.='13.0']" // 13.00 * (cite_read_boost + aqp.classic_scoring.modifier) 
 				);
 		assertQ(req("q", "author:\"Einstein, A\" AND author:\"Anders\"",
 		    "aqp.constant_scoring", "author^13",
+		    "aqp.classic_scoring.modifier", "0.48",
         "fl", "recid,score"),
 				"//*[@numFound='1']",
 				"//doc/int[@name='recid'][.='100']",
 				"//doc/float[@name='score'][.='26.0']");
 		assertQ(req("q", "author:\"Einstein, A\" OR author:\"Anders\"",
         "aqp.constant_scoring", "author^13",
+        "aqp.classic_scoring.modifier", "0.48",
         "fl", "recid,score"),
         "//*[@numFound='1']",
         "//doc/int[@name='recid'][.='100']",

diff --git a/contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeFulltextParsing.java b/contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeFulltextParsing.java
@@ -1222,12 +1222,16 @@ public void testOtherCases() throws Exception {
     assertQ(req("q", "title:\"GBT Survey of 50 Faint Fermi\"~2"), 
         "//*[@numFound>='4']");
 
-    assertQ(req("q", "title:\"A 350-MHz GBT Survey of 50 Faint Fermi γ-ray Sources for Radio Millisecond Pulsars\""), 
+
+    assertQ(req("q", "title:\"A 350-MHz GBT Survey of 50 Faint Fermi γ-ray Sources for Radio Millisecond Pulsars\"",
+        "indent", "true",
+        "debugQuery", "true"), 
         "//*[@numFound='4']",
         "//doc/str[@name='id'][.='400']",
         "//doc/str[@name='id'][.='401']",
         "//doc/str[@name='id'][.='402']",
-        "//doc/str[@name='id'][.='403']");
+        "//doc/str[@name='id'][.='403']"
+        );
     assertQ(req("q", "title:\"A 350-MHz GBT Survey of 50 Faint Fermi γ ray Sources for Radio Millisecond Pulsars\""), 
         "//*[@numFound='4']",
         "//doc/str[@name='id'][.='400']",

diff --git a/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java b/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java
@@ -138,6 +138,7 @@ public void tearDown() throws Exception {
 
     public void testUnfieldedSearch() throws Exception {
 
+
       // have constant scoring work even for unfielded searches
       assertQueryEquals(req("defType", "aqp", "q", "foo bar",
           "qf", "bibcode^5 title^10",
@@ -375,6 +376,9 @@ public void testSpecialCases() throws Exception {
                   "title", "title bitle"));
       assertU(commit("waitSearcher", "true"));
 
+      assertQ(req("q", "similar(foo bar baz title bitle, input abstract title, 100, 100, 1, 1)"),
+          "//*[@numFound='1']",
+          "//doc/str[@name='id'][.='2']");
 
       // similar()
       assertQueryEquals(req("defType", "aqp", "q", "similar(foo bar baz, input)"),

diff --git a/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml b/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml
@@ -319,14 +319,23 @@
        Modified qf
        original: first_author^3.0 author^2 title^1.4 abstract^1.3 keyword^1.4 keyword_norm^1.4 all body^0.1 year
        new: first_author^5 author^2 title^1.5 abstract^1.3 identifier^1 bibstem^1 year^2
+       
+       August 28, 2019
+       Modified qf:
+       old: first_author^5 author^2 title^1.5 abstract^1.3 identifier^1 bibstem^1 year^2
+       new: first_author^0.9 author^0.85 year^0.8 title^0.8 abstract^0.7 identifier^0.8 bibstem^0.8 keyword^0.8
         -->
-       <str name="qf">first_author^5 author^2 title^1.5 abstract^1.3 identifier^1 bibstem^1 year^2</str>
+       <str name="qf">first_author^0.9 author^0.85 year^0.8 title^0.8 abstract^0.7 identifier^0.8 bibstem^0.8 keyword^0.8</str>
+
        <str name="defType">aqp</str>
        <str name="aqp.unfielded.tokens.strategy">disjuncts</str>
        <str name="aqp.unfielded.tokens.new.type">simple</str>
        <str name="aqp.unfielded.ignore.fields">entdate pubdate year</str>
        <str name="aqp.unfielded.tokens.function.name">edismax_combined_aqp</str>
        <str name="useFastVectorHighlighter">true</str>
+       <str name="aqp.classic_scoring.modifier">0.5</str>
+       <str name="aqp.constant_scoring">first_author^14 author^13 year^10 bibstem^10</str>
+       <str name="aqp.multiphrase.keep_one">SYNONYM</str>
      </lst>
      <lst name="invariants">
        <str name="echoParams">explicit</str>
@@ -349,6 +358,9 @@
        <str name="aqp.unfielded.ignore.fields">entdate pubdate year</str>
        <str name="aqp.unfielded.tokens.function.name">edismax_combined_aqp</str>
        <str name="useFastVectorHighlighter">true</str>
+       <str name="aqp.classic_scoring.modifier">0.5</str>
+       <str name="aqp.constant_scoring">first_author^14 author^13 year^10 bibstem^10</str>
+       <str name="aqp.multiphrase.keep_one">SYNONYM</str>
      </lst>
     <lst name="invariants">
        <str name="echoParams">explicit</str>
@@ -697,7 +709,7 @@
        <str name="aqp.floatFields">cite_read_boost,citation_count_norm</str>
        <str name="aqp.intFields">recid,pubdate_sort,citation_count,classic_factor,simbid,read_count,author_count,page_count,data_count</str>
        <str name="aqp.authorFields">author,first_author,book_author,editor</str>
-       <str name="aqp.humanized.dates">pubdate:date,entdate:entry_date</str>
+       <str name="aqp.humanized.dates">pubdate:date,entdate:entry_date:timestamp</str>
        <str name="aqp.force.fuzzy.phrases">author,first_author,book_author,editor</str>
     </lst>
   </queryParser>