Neo4j节点索引：如何更改Lucene的默认相似度？

Question

I'm using Neo4j (1.9) and lucene core 3.5, through the support offered by the neo4j-lucene-index library. In my code, I create a new node index like that:

    HashMap<String, String> stringMap = new HashMap<String, String>();
    stringMap.put("type", "fulltext");
    stringMap.put(IndexManager.PROVIDER, "lucene");
    stringMap.put("to_lower_case", "false");
    stringMap.put("similarity", MySimilarity.class.getName());
    stringMap.put("analyzer", MyAnalyzer.class.getName());
    simpleIndex = (LuceneIndex<Node>) template.getGraphDatabaseService()
            .index().forNodes("simple-index", stringMap);

and query the node index by means of a multi-field query:

    Analyzer analyzer = new MyAnalyzer();
    Query query = null;
    try {
        query = MultiFieldQueryParser.parse(Version.LUCENE_35,
                queriesArray, fieldsArray, flagsArray, analyzer);
    } catch (ParseException e) {
        e.printStackTrace();
        return;
    }
    if (query == null)
        return;
    QueryContext qc = new QueryContext(query).sortByScore();
    IndexHits<Node> hits = simpleIndex.query(qc);

How can I change the Similarity implementation, used for sorting by score, to MySimilarity class (which extends the DefaultSimilarity and overrides some of its methods)? The reported query always uses the DefaultSimilarity and I was not able to find any hook to set the similarity used by the searcher activated during my query.

Answer 1

The fact that the supplied similarity isn't used during queries is a bug and should be fixed. Created https://github.com/neo4j/neo4j/issues/375 to track it.