Examples of org.sindice.siren.search.node.NodeFuzzyQuery

org.sindice.siren.search.node.NodeFuzzyQuery
Implements the fuzzy search query. The similarity measurement is based on the Levenshtein (edit distance) algorithm.
This query uses {@link MultiTermQuery.TopTermsScoringBooleanQueryRewrite}as default. So terms will be collected and scored according to their edit distance. Only the top terms are used for building the {@link NodeBooleanQuery}. It is not recommended to change the rewrite mode for fuzzy queries.
Code taken from {@link FuzzyQuery} and adapted for SIREn.

  }


  @Test
  public void testFuzzyQuery2()
  throws Exception {
    final NodeQuery q1 = new NodeFuzzyQuery(new Term(SirenTestCase.DEFAULT_TEST_FIELD, "michel"));
    this._assertSirenQuery(new LuceneProxyNodeQuery(q1), "michel~");


    final TwigQuery q2 = new TwigQuery(1);
    q2.addChild(q1, NodeBooleanClause.Occur.MUST);
    this._assertSirenQuery(new LuceneProxyNodeQuery(q2), "* : michel~");


    final int numEdits = FuzzyQuery.floatToEdits(0.8f, "michel".codePointCount(0, "michel".length()));
    final NodeQuery q3 = new NodeFuzzyQuery(new Term(SirenTestCase.DEFAULT_TEST_FIELD, "michel"), numEdits);
    this._assertSirenQuery(new LuceneProxyNodeQuery(q3), "michel~0.8");


    // first tilde is escaped, not the second one
    final NodeQuery q4 = new NodeFuzzyQuery(new Term(SirenTestCase.DEFAULT_TEST_FIELD, "http://sw.deri.org/~aida"));
    this._assertSirenQuery(new LuceneProxyNodeQuery(q4), "'http://sw.deri.org/~aida'~");
  }

View Full Code Here

    this.addDocument("<aabbb>");
    this.addDocument("<abbbb>");
    this.addDocument("<bbbbb>");
    this.addDocument("<ddddd>");


    LuceneProxyNodeQuery query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 0));
    ScoreDoc[] hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);


    // same with prefix
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 1));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 2));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 3));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 4));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(2, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 5));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 6));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);


    // test scoring
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "bbbbb"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals("3 documents should match", 3, hits.length);
    List<Integer> order = Arrays.asList(5,4,3);
    for (int i = 0; i < hits.length; i++) {
      assertEquals((int) order.get(i), hits[i].doc);
    }


    // test pq size by supplying maxExpansions=2
    // This query would normally return 3 documents, because 3 terms match (see above):
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "bbbbb"), NodeFuzzyQuery.defaultMaxEdits, 0, 2, false));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals("only 2 documents should match", 2, hits.length);
    order = Arrays.asList(5,4);
    for (int i = 0; i < hits.length; i++) {
      assertEquals((int) order.get(i), hits[i].doc);
    }


    // not similar enough:
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "xxxxx"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(0, hits.length);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaccc"), NodeFuzzyQuery.defaultMaxEdits, 0));   // edit distance to "aaaaa" = 3
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(0, hits.length);


    // query identical to a word in the index:
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaaa"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    assertEquals(0, hits[0].doc);
    // default allows for up to two edits:
    assertEquals(1, hits[1].doc);
    assertEquals(2, hits[2].doc);


    // query similar to a word in the index:
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    assertEquals(0, hits[0].doc);
    assertEquals(1, hits[1].doc);
    assertEquals(2, hits[2].doc);


    // now with prefix
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 1));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    assertEquals(0, hits[0].doc);
    assertEquals(1, hits[1].doc);
    assertEquals(2, hits[2].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 2));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    assertEquals(0, hits[0].doc);
    assertEquals(1, hits[1].doc);
    assertEquals(2, hits[2].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 3));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(3, hits.length);
    assertEquals(0, hits[0].doc);
    assertEquals(1, hits[1].doc);
    assertEquals(2, hits[2].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 4));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(2, hits.length);
    assertEquals(0, hits[0].doc);
    assertEquals(1, hits[1].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "aaaac"), NodeFuzzyQuery.defaultMaxEdits, 5));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(0, hits.length);




    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    assertEquals(6, hits[0].doc);


    // now with prefix
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 1));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    assertEquals(6, hits[0].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 2));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    assertEquals(6, hits[0].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 3));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    assertEquals(6, hits[0].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 4));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(1, hits.length);
    assertEquals(6, hits[0].doc);
    query = dq(new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 5));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(0, hits.length);




    // different field = no match:
    query = dq(new NodeFuzzyQuery(new Term("anotherfield", "ddddX"), NodeFuzzyQuery.defaultMaxEdits, 0));
    hits = searcher.search(query, null, 1000).scoreDocs;
    assertEquals(0, hits.length);
  }

View Full Code Here

  public void testBoostOnlyRewrite() throws Exception {
    this.addDocument("<Lucene>");
    this.addDocument("<Lucene>");
    this.addDocument("<Lucenne>");


    final NodeFuzzyQuery query = new NodeFuzzyQuery(new Term(DEFAULT_TEST_FIELD, "Lucene"));
    query.setRewriteMethod(new MultiNodeTermQuery.TopTermsBoostOnlyNodeBooleanQueryRewrite(50));
    final ScoreDoc[] hits = searcher.search(dq(query), null, 1000).scoreDocs;
    assertEquals(3, hits.length);
  }

View Full Code Here

    String text = fuzzyNode.getTextAsString();
    
    int numEdits = FuzzyQuery.floatToEdits(fuzzyNode.getSimilarity(), 
        text.codePointCount(0, text.length()));


    return new NodeFuzzyQuery(new Term(fuzzyNode.getFieldAsString(), fuzzyNode
        .getTextAsString()), numEdits, fuzzyNode
        .getPrefixLength());
  }

View Full Code Here

TOP

Related Classes of org.sindice.siren.search.node.NodeFuzzyQuery

org.apache.lucene.index.SingleTermsEnum

org.apache.lucene.search.FuzzyTermsEnum

org.sindice.siren.qparser.keyword.builders.NodeFuzzyQueryNodeBuilder

org.sindice.siren.qparser.keyword.KeywordQueryParserTest

org.sindice.siren.search.node.TestNodeFuzzyQuery

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.