使用Lucene使多值字段评分了解字段值

Question

I am trying to associate a list of (multi words) tags to each Document. So for each document I am adding multiple StringField entries with "tag" as fieldName.

When searching, I am expecting the score to be proportional to the ratio of tags I am successfully matching, for instance:

• 0.5 if I match half of the tags
• 1.0 if I match them all.

But it seems that the number of tags is not taken into consideration in the score.

When testing on these four documents:

 - tags.put("doc1", "piano, electric guitar, violon");

 - tags.put("doc2", "piano, electric guitar");

 - tags.put("doc3", "piano");

 - tags.put("doc4", "electric guitar"); 

What I get is:

 - Score : 1.0 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc4> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar>>
 - Score : 1.0 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc2> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:piano> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar>>
 - Score : 1.0 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc1> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:piano> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:violon>>

How can I change this behavior? Am I missing the right way of doing things?

Below is my test code.

Best regards,

Renaud

public class LuceneQueryTest {

    Analyzer analyzer;
    BasicIndex basicIndex;
    LinkedList<String> phrases;
    Query query;
    Map<Document, Float> results;

    @Test
    public void testListOfTags() throws Exception {

        analyzer = new StandardAnalyzer();

        basicIndex = new BasicIndex(analyzer);

        Map<String, String> tags = new HashMap();

        tags.put("doc1", "piano, electric guitar, violon");

        tags.put("doc2", "piano, electric guitar");

        tags.put("doc3", "piano");

        tags.put("doc4", "electric guitar");

        Queue<String> queue = new LinkedList<>();
        queue.addAll(tags.keySet());

        basicIndex.index(new Supplier<Document>() {

            public Document get() {
                Document doc = new Document();

                if (queue.isEmpty()) {
                    return null;
                }

                String docName = queue.poll();

                System.out.println("**** "+docName);

                String tag = tags.get(docName);
                doc.add(new StringField("id", docName, Field.Store.YES));

                for (String tagItem : tag.split("\\,")) {
                    System.out.println(tagItem);
                    Field tagField;
                    tagField = new StringField("tag",tagItem,Field.Store.YES);

                    System.out.println(tagField);

                    doc.add(tagField);
                }
                return doc;
            }
        });

        BooleanQuery booleanQuery = new BooleanQuery();
        //booleanQuery.add(new TermQuery(new Term("tag", "piano")), BooleanClause.Occur.SHOULD);
        booleanQuery.add(new TermQuery(new Term("tag", "electric guitar")), BooleanClause.Occur.SHOULD);

        //Query parsedQuery = new QueryParser("tag", analyzer).parse("tag:\"electric guitar\"");
        query = booleanQuery;
        //query = parsedQuery;


        System.out.println(query);

        results = basicIndex.search(query);
        displayResults(results);

        System.out.println(Arrays.toString(basicIndex.document(3).getValues("tag")));

    }

    private void displayResults(Map<Document, Float> results) {
        results.forEach((Document doc, Float score) -> {
            System.out.println("Score : " + score + " \n Doc : " + doc);
        });
    }
}

Code for the BasicIndex (test utility) class:

import java.io.IOException;
import java.util.LinkedHashMap;
import java.util.LinkedList;
import java.util.Map;
import java.util.function.Function;
import java.util.function.Supplier;
import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.Field;
import org.apache.lucene.document.TextField;
import org.apache.lucene.index.DirectoryReader;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.IndexWriterConfig;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.ScoreDoc;
import org.apache.lucene.search.TopScoreDocCollector;
import org.apache.lucene.store.Directory;
import org.apache.lucene.store.RAMDirectory;
import org.apache.lucene.util.Version;

/**
 *
 * @author renaud
 */
public class BasicIndex {

final Directory directory = new RAMDirectory();
final IndexWriter indexWriter;
final Analyzer analyzer;

public BasicIndex(Analyzer analyzer) {
    this.analyzer = analyzer;
    this.indexWriter = newIndexWriter();
}

public Analyzer getAnalyzer() {
    return analyzer;
}

private IndexWriter newIndexWriter() {
    IndexWriterConfig config = new IndexWriterConfig(Version.LATEST, analyzer);
    try {
        return new IndexWriter(directory, config);
    } catch (IOException ex) {
        throw new RuntimeException(ex);
    }
}

public IndexSearcher newIndexSearcher() {
    return new IndexSearcher(newIndexReader());
}

public IndexReader newIndexReader() {
    IndexReader reader;
    try {

        reader = DirectoryReader.open(directory);
    } catch (IOException ex) {
        throw ExceptionUtils.asRuntimeException(ex);
    }
    return reader;
}

public void index(LinkedList<String> phrases, final String fieldName) {
    index(phrases, (String phrase) -> {
        Document doc = new Document();

        Field workField = new TextField(fieldName, phrase, Field.Store.YES);
        doc.add(workField);
        return doc;
    });
}

public void index(Supplier<Document> documents) {
    Document document;
    while ((document = documents.get()) != null) {
        try {
            indexWriter.addDocument(document);
        } catch (IOException e) {
            throw ExceptionUtils.asRuntimeException(e);
        }
    }
    close();
}

public void index(LinkedList<String> phrases, Function<String, Document> docBuilder) {
    for (String phrase : phrases) {
        try {
            indexWriter.addDocument(docBuilder.apply(phrase));
        } catch (IOException e) {
            throw ExceptionUtils.asRuntimeException(e);
        }
    }
    close();
}

private void close() {
    IOUtils.closeSilently(indexWriter);
}

public Map<Document, Float> search(Query query) {
    final IndexSearcher indexSearcher = newIndexSearcher();
    int hitsPerPage = 10;
    TopScoreDocCollector collector = TopScoreDocCollector.create(hitsPerPage, true);
    try {
        indexSearcher.search(query, collector);
    } catch (IOException ex) {
        throw new RuntimeException(ex);
    }

    ScoreDoc[] hits = collector.topDocs().scoreDocs;

    Map<Document, Float> results = new LinkedHashMap<>();
    for (int i = 0; i < hits.length; ++i) {
        ScoreDoc scoreDoc = hits[i];
        int docId = scoreDoc.doc;
        float score = scoreDoc.score;
        Document doc;
        try {
            doc = indexSearcher.doc(docId);
        } catch (IOException ex) {
            throw new RuntimeException(ex);
        }
        results.put(doc, score);
    }


    return results;
}

public Document document(int i){
    try {
        return newIndexSearcher().doc(i);
    } catch (IOException ex) {
        throw new RuntimeException(ex);
    }
}

}

Answer 1

Ok the solution I finally came to is:

1. Add an IntField containing the number of tags of the doc
2. Use a BoostedQuery using a Reciprocal function of the IntFieldSource

Also I found out that SOLR is a good keyword for searching info about Lucene over the internet as it is a lot more documented while staying close to the java code.

I am pretty happy with the result:

Score : 0.5 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc4> stored<count:1> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar>>
Score : 0.33333334 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc2> stored<count:2> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:piano> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar>>
Score : 0.25 
 Doc : Document<stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<id:doc1> stored<count:3> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:piano> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:electric guitar> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<tag:violon>>

And the updated code:

    @Test
    public void testListOfTags() throws Exception {

        analyzer = new StandardAnalyzer();

        basicIndex = new BasicIndex(analyzer);

        Map<String, String> tags = new HashMap();

        tags.put("doc1", "piano, electric guitar, violon");

        tags.put("doc2", "piano, electric guitar");

        tags.put("doc3", "piano");

        tags.put("doc4", "electric guitar");

        Queue<String> queue = new LinkedList<>();
        queue.addAll(tags.keySet());

        basicIndex.index(new Supplier<Document>() {

            public Document get() {
                Document doc = new Document();

                if (queue.isEmpty()) {
                    return null;
                }

                String docName = queue.poll();

                System.out.println("**** " + docName);

                String tag = tags.get(docName);
                doc.add(new StringField("id", docName, Field.Store.YES));
                String[] tags = tag.split("\\,");

                Field tagCountField = new IntField("count", tags.length, Field.Store.YES);
                doc.add(tagCountField);

                for (String tagItem : tags) {
                    System.out.println(tagItem);
                    Field tagField;
                    tagField = new StringField("tag", tagItem.trim(), Field.Store.YES);

                    System.out.println(tagField);

                    doc.add(tagField);
                }
                return doc;
            }
        });

        BooleanQuery booleanQuery = new BooleanQuery();
        //booleanQuery.add(new TermQuery(new Term("tag", "piano")), BooleanClause.Occur.SHOULD);
        booleanQuery.add(new TermQuery(new Term("tag", "electric guitar")), BooleanClause.Occur.SHOULD);

        //Query parsedQuery = new QueryParser("tag", analyzer).parse("tag:\"electric guitar\"");
        query = booleanQuery;
        //query = parsedQuery;


        ValueSource boostSource = new ReciprocalFloatFunction(new IntFieldSource("count"), 1, 1, 1);
        query = new BoostedQuery(query, boostSource);

        System.out.println(query);

        results = basicIndex.search(query);
        displayResults(results);

        System.out.println(Arrays.toString(basicIndex.document(3).getValues("tag")));

    }