Search in sources :

Example 11 with ClassicSimilarity

use of org.apache.lucene.search.similarities.ClassicSimilarity in project lucene-solr by apache.

the class TestClassicSimilarityFactory method testParams.

/** Classic w/ explicit params */
public void testParams() throws Exception {
    ClassicSimilarity sim = getSimilarity("text_overlap", ClassicSimilarity.class);
    assertEquals(false, sim.getDiscountOverlaps());
}
Also used : ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity)

Example 12 with ClassicSimilarity

use of org.apache.lucene.search.similarities.ClassicSimilarity in project lucene-solr by apache.

the class TestClassicSimilarityFactory method testDefaults.

/** Classic w/ default parameters */
public void testDefaults() throws Exception {
    ClassicSimilarity sim = getSimilarity("text", ClassicSimilarity.class);
    assertEquals(true, sim.getDiscountOverlaps());
}
Also used : ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity)

Example 13 with ClassicSimilarity

use of org.apache.lucene.search.similarities.ClassicSimilarity in project lucene-solr by apache.

the class SweetSpotSimilarityTest method testSweetSpotComputeNorm.

public void testSweetSpotComputeNorm() throws IOException {
    final SweetSpotSimilarity ss = new SweetSpotSimilarity();
    ss.setLengthNormFactors(1, 1, 0.5f, true);
    Similarity d = new ClassicSimilarity();
    Similarity s = ss;
    // base case, should degrade
    for (int i = 1; i < 1000; i++) {
        assertEquals("base case: i=" + i, computeNorm(d, "bogus", i), computeNorm(s, "bogus", i), 0.0f);
    }
    // make a sweet spot
    ss.setLengthNormFactors(3, 10, 0.5f, true);
    for (int i = 3; i <= 10; i++) {
        assertEquals("3,10: spot i=" + i, 1.0f, computeNorm(ss, "bogus", i), 0.0f);
    }
    for (int i = 10; i < 1000; i++) {
        final float normD = computeNorm(d, "bogus", i - 9);
        final float normS = computeNorm(s, "bogus", i);
        assertEquals("3,10: 10<x : i=" + i, normD, normS, 0.01f);
    }
    // separate sweet spot for certain fields
    final SweetSpotSimilarity ssBar = new SweetSpotSimilarity();
    ssBar.setLengthNormFactors(8, 13, 0.5f, false);
    final SweetSpotSimilarity ssYak = new SweetSpotSimilarity();
    ssYak.setLengthNormFactors(6, 9, 0.5f, false);
    final SweetSpotSimilarity ssA = new SweetSpotSimilarity();
    ssA.setLengthNormFactors(5, 8, 0.5f, false);
    final SweetSpotSimilarity ssB = new SweetSpotSimilarity();
    ssB.setLengthNormFactors(5, 8, 0.1f, false);
    Similarity sp = new PerFieldSimilarityWrapper() {

        @Override
        public Similarity get(String field) {
            if (field.equals("bar"))
                return ssBar;
            else if (field.equals("yak"))
                return ssYak;
            else if (field.equals("a"))
                return ssA;
            else if (field.equals("b"))
                return ssB;
            else
                return ss;
        }
    };
    for (int i = 3; i <= 10; i++) {
        assertEquals("f: 3,10: spot i=" + i, 1.0f, computeNorm(sp, "foo", i), 0.0f);
    }
    for (int i = 10; i < 1000; i++) {
        final float normD = computeNorm(d, "foo", i - 9);
        final float normS = computeNorm(sp, "foo", i);
        assertEquals("f: 3,10: 10<x : i=" + i, normD, normS, 0.01f);
    }
    for (int i = 8; i <= 13; i++) {
        assertEquals("f: 8,13: spot i=" + i, 1.0f, computeNorm(sp, "bar", i), 0.01f);
    }
    for (int i = 6; i <= 9; i++) {
        assertEquals("f: 6,9: spot i=" + i, 1.0f, computeNorm(sp, "yak", i), 0.01f);
    }
    for (int i = 13; i < 1000; i++) {
        final float normD = computeNorm(d, "bar", i - 12);
        final float normS = computeNorm(sp, "bar", i);
        assertEquals("f: 8,13: 13<x : i=" + i, normD, normS, 0.01f);
    }
    for (int i = 9; i < 1000; i++) {
        final float normD = computeNorm(d, "yak", i - 8);
        final float normS = computeNorm(sp, "yak", i);
        assertEquals("f: 6,9: 9<x : i=" + i, normD, normS, 0.01f);
    }
    for (int i = 9; i < 1000; i++) {
        final float normSS = computeNorm(sp, "a", i);
        final float normS = computeNorm(sp, "b", i);
        assertTrue("s: i=" + i + " : a=" + normSS + " < b=" + normS, normSS < normS);
    }
}
Also used : ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity) ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity) TFIDFSimilarity(org.apache.lucene.search.similarities.TFIDFSimilarity) Similarity(org.apache.lucene.search.similarities.Similarity) PerFieldSimilarityWrapper(org.apache.lucene.search.similarities.PerFieldSimilarityWrapper)

Example 14 with ClassicSimilarity

use of org.apache.lucene.search.similarities.ClassicSimilarity in project lucene-solr by apache.

the class SweetSpotSimilarityTest method testSweetSpotTf.

public void testSweetSpotTf() {
    SweetSpotSimilarity ss = new SweetSpotSimilarity();
    TFIDFSimilarity d = new ClassicSimilarity();
    TFIDFSimilarity s = ss;
    // tf equal
    ss.setBaselineTfFactors(0.0f, 0.0f);
    for (int i = 1; i < 1000; i++) {
        assertEquals("tf: i=" + i, d.tf(i), s.tf(i), 0.0f);
    }
    // tf higher
    ss.setBaselineTfFactors(1.0f, 0.0f);
    for (int i = 1; i < 1000; i++) {
        assertTrue("tf: i=" + i + " : d=" + d.tf(i) + " < s=" + s.tf(i), d.tf(i) < s.tf(i));
    }
    // tf flat
    ss.setBaselineTfFactors(1.0f, 6.0f);
    for (int i = 1; i <= 6; i++) {
        assertEquals("tf flat1: i=" + i, 1.0f, s.tf(i), 0.0f);
    }
    ss.setBaselineTfFactors(2.0f, 6.0f);
    for (int i = 1; i <= 6; i++) {
        assertEquals("tf flat2: i=" + i, 2.0f, s.tf(i), 0.0f);
    }
    for (int i = 6; i <= 1000; i++) {
        assertTrue("tf: i=" + i + " : s=" + s.tf(i) + " < d=" + d.tf(i), s.tf(i) < d.tf(i));
    }
    // stupidity
    assertEquals("tf zero", 0.0f, s.tf(0), 0.0f);
}
Also used : ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity) TFIDFSimilarity(org.apache.lucene.search.similarities.TFIDFSimilarity)

Example 15 with ClassicSimilarity

use of org.apache.lucene.search.similarities.ClassicSimilarity in project lucene-solr by apache.

the class TestTaxonomyFacetCounts method testReallyNoNormsForDrillDown.

public void testReallyNoNormsForDrillDown() throws Exception {
    Directory dir = newDirectory();
    Directory taxoDir = newDirectory();
    IndexWriterConfig iwc = newIndexWriterConfig(new MockAnalyzer(random()));
    iwc.setSimilarity(new PerFieldSimilarityWrapper() {

        final Similarity sim = new ClassicSimilarity();

        @Override
        public Similarity get(String name) {
            assertEquals("field", name);
            return sim;
        }
    });
    TaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode.CREATE);
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir, iwc);
    FacetsConfig config = new FacetsConfig();
    Document doc = new Document();
    doc.add(newTextField("field", "text", Field.Store.NO));
    doc.add(new FacetField("a", "path"));
    writer.addDocument(config.build(taxoWriter, doc));
    writer.close();
    IOUtils.close(taxoWriter, dir, taxoDir);
}
Also used : ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity) Similarity(org.apache.lucene.search.similarities.Similarity) ClassicSimilarity(org.apache.lucene.search.similarities.ClassicSimilarity) FacetsConfig(org.apache.lucene.facet.FacetsConfig) FacetField(org.apache.lucene.facet.FacetField) Document(org.apache.lucene.document.Document) DirectoryTaxonomyWriter(org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyWriter) DirectoryTaxonomyWriter(org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyWriter) MockAnalyzer(org.apache.lucene.analysis.MockAnalyzer) PerFieldSimilarityWrapper(org.apache.lucene.search.similarities.PerFieldSimilarityWrapper) RandomIndexWriter(org.apache.lucene.index.RandomIndexWriter) Directory(org.apache.lucene.store.Directory) IndexWriterConfig(org.apache.lucene.index.IndexWriterConfig)

Aggregations

ClassicSimilarity (org.apache.lucene.search.similarities.ClassicSimilarity)43 RandomIndexWriter (org.apache.lucene.index.RandomIndexWriter)14 Document (org.apache.lucene.document.Document)13 Term (org.apache.lucene.index.Term)12 Directory (org.apache.lucene.store.Directory)10 IndexReader (org.apache.lucene.index.IndexReader)9 Similarity (org.apache.lucene.search.similarities.Similarity)9 TermQuery (org.apache.lucene.search.TermQuery)7 BM25Similarity (org.apache.lucene.search.similarities.BM25Similarity)7 MockAnalyzer (org.apache.lucene.analysis.MockAnalyzer)6 ConstValueSource (org.apache.lucene.queries.function.valuesource.ConstValueSource)5 DocFreqValueSource (org.apache.lucene.queries.function.valuesource.DocFreqValueSource)4 DoubleConstValueSource (org.apache.lucene.queries.function.valuesource.DoubleConstValueSource)4 IDFValueSource (org.apache.lucene.queries.function.valuesource.IDFValueSource)4 JoinDocFreqValueSource (org.apache.lucene.queries.function.valuesource.JoinDocFreqValueSource)4 LiteralValueSource (org.apache.lucene.queries.function.valuesource.LiteralValueSource)4 MaxDocValueSource (org.apache.lucene.queries.function.valuesource.MaxDocValueSource)4 IndexSearcher (org.apache.lucene.search.IndexSearcher)4 Query (org.apache.lucene.search.Query)4 IndexWriterConfig (org.apache.lucene.index.IndexWriterConfig)3