Uploaded image for project: 'Lucene - Core'
  1. Lucene - Core
  2. LUCENE-6789

change IndexSearcher default similarity to BM25



    • Improvement
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • None
    • 6.0
    • None
    • None
    • New


      Since Lucene 4.0, the statistics needed for this are always present, so we can make the change without any degradation.

      I think the change should be a 6.0 change only: it will prevent any surprises. DefaultSimilarity is renamed to ClassicSimilarity to prevent confusion. No indexing change is needed as we use the same norm format, its just a runtime switch. Users can just do IndexSearcher.setSimilarity(new ClassicSimilarity()) to get the old behavior. I did not change solr's default here, I think that should be a separate issue, since it has more concerns: e.g. factories in configuration files and so on.

      One issue was the generation of synonym queries (posinc=0) by QueryBuilder (used by parsers). This is kind of a corner case (query-time synonyms), but we should make it nicer. The current code in trunk disables coord, which makes no sense for anything but the vector space impl. Instead, this patch adds a SynonymQuery which treats occurrences of any term as a single pseudoterm. With english wordnet as a query-time synonym dict, this query gives 12% improvement in MAP for title queries on BM25, and 2% with Classic (not significant). So its a better generic approach for synonyms that works with all scoring models.

      I wanted to use BlendedTermQuery, but it seems to have problems at a glance, it tries to "take on the world", it has problems like not working with distributed scoring (doesn't consult indexsearcher for stats). Anyway this one is a different, simpler approach, which only works for a single field, and which calls tf(sum) a single time.


        1. LUCENE-6789.patch
          144 kB
          Robert Muir

        Issue Links



              Unassigned Unassigned
              rcmuir Robert Muir
              0 Vote for this issue
              5 Start watching this issue