| Index: trunk/WikiWord/WikiWord/src/test/java/de/brightbyte/wikiword/disambig/CoherenceDisambiguatorTest.java |
| — | — | @@ -6,10 +6,13 @@ |
| 7 | 7 | import java.util.List; |
| 8 | 8 | import java.util.Map; |
| 9 | 9 | |
| | 10 | +import de.brightbyte.data.LabeledMatrix; |
| | 11 | +import de.brightbyte.data.MapLabeledMatrix; |
| 10 | 12 | import de.brightbyte.data.Pair; |
| 11 | 13 | import de.brightbyte.io.ConsoleIO; |
| 12 | 14 | import de.brightbyte.io.Output; |
| 13 | 15 | import de.brightbyte.util.PersistenceException; |
| | 16 | +import de.brightbyte.wikiword.disambig.CoherenceDisambiguator.CoherenceDisambiguation; |
| 14 | 17 | import de.brightbyte.wikiword.disambig.Disambiguator.Interpretation; |
| 15 | 18 | import de.brightbyte.wikiword.disambig.Disambiguator.Disambiguation; |
| 16 | 19 | import de.brightbyte.wikiword.model.LocalConcept; |
| — | — | @@ -24,6 +27,33 @@ |
| 25 | 28 | super(); |
| 26 | 29 | } |
| 27 | 30 | |
| | 31 | + public void testGetScore() throws PersistenceException { |
| | 32 | + CoherenceDisambiguator disambiguator = new CoherenceDisambiguator(meaningFetcher, featureFetcher, 10); |
| | 33 | + |
| | 34 | + LabeledMatrix<LocalConcept, LocalConcept> similarities = new MapLabeledMatrix<LocalConcept, LocalConcept>(true); |
| | 35 | + |
| | 36 | + LocalConcept city_of_London = getConcept("City_of_London"); |
| | 37 | + LocalConcept united_Kingdom = getConcept("United_Kingdom"); |
| | 38 | + |
| | 39 | + //united_Kingdom.setCardinality(100000); |
| | 40 | + |
| | 41 | + Pair<Term, LocalConcept> uk_as_United_Kingdom = new Pair<Term, LocalConcept>(new Term("UK"), united_Kingdom); |
| | 42 | + Pair<Term, LocalConcept> london_as_City_of_London = new Pair<Term, LocalConcept>(new Term("London"), city_of_London); |
| | 43 | + |
| | 44 | + CoherenceDisambiguator.Interpretation interp = new CoherenceDisambiguator.Interpretation(uk_as_United_Kingdom, london_as_City_of_London); |
| | 45 | + CoherenceDisambiguation r1 = disambiguator.getScore(interp, null, similarities, featureFetcher); |
| | 46 | + |
| | 47 | + int oldPop = city_of_London.getCardinality(); |
| | 48 | + city_of_London.setCardinality(oldPop*2); |
| | 49 | + |
| | 50 | + CoherenceDisambiguation r2 = disambiguator.getScore(interp, null, similarities, featureFetcher); |
| | 51 | + city_of_London.setCardinality(oldPop); |
| | 52 | + |
| | 53 | + double score1 = r1.getScore(); |
| | 54 | + double score2 = r2.getScore(); |
| | 55 | + assertTrue("More popularity implies better score", score1 < score2 ); |
| | 56 | + } |
| | 57 | + |
| 28 | 58 | public void testGetSequenceInterpretations() throws PersistenceException { |
| 29 | 59 | CoherenceDisambiguator disambiguator = new CoherenceDisambiguator(meaningFetcher, featureFetcher, 10); |
| 30 | 60 | |
| Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/disambig/CoherenceDisambiguator.java |
| — | — | @@ -77,7 +77,10 @@ |
| 78 | 78 | if (pop<0.5) return 0; |
| 79 | 79 | if (pop<1) pop=1; |
| 80 | 80 | |
| 81 | | - double n = 1 - 1/(Math.sqrt(Math.log(pop))+1); //XXX: black voodoo magic ad hoc formula with no deeper meaing. |
| | 81 | + //XXX: black voodoo magic ad hoc formula with no deeper meaing. |
| | 82 | + //double n = 1 - 1/(Math.log(pop)+1); //normalized log scale |
| | 83 | + //double n = 1 - 1/(Math.sqrt(Math.log(pop))+1); //dampened normalized log scale |
| | 84 | + double n =1 - (0.5/Math.sqrt((pop+200)/200)); //nice and smooth, but has magic params that may depend on the wiki |
| 82 | 85 | return n; |
| 83 | 86 | } |
| 84 | 87 | }; |
| — | — | @@ -436,7 +439,8 @@ |
| 437 | 440 | |
| 438 | 441 | LabeledVector<Integer> sum = ConceptFeatures.newIntFeaturVector( concepts.size() * 200 ); //XXX: magic number |
| 439 | 442 | Map<Integer, ConceptFeatures<C, Integer>> disambigFeatures = new HashMap<Integer, ConceptFeatures<C, Integer>>(); |
| 440 | | - double sim = 0, pop = 0, weight = 0; |
| | 443 | + double sim = 0, pop = 0, weight = 0, popf = 0, simf = 0; |
| | 444 | + |
| 441 | 445 | int i=0, j=0; |
| 442 | 446 | for (Map.Entry<TermReference, C> ea: concepts.entrySet()) { |
| 443 | 447 | C a = ea.getValue(); |
| — | — | @@ -496,6 +500,8 @@ |
| 497 | 501 | d = doubleSanity(d, "normal similarity score for "+a+" / "+b, "check similarityMeasure!", 0, 0.1, 1, 0.1); |
| 498 | 502 | |
| 499 | 503 | sim += d; |
| | 504 | + simf += similarityNormalizer.apply(d); |
| | 505 | + |
| 500 | 506 | simCount ++; |
| 501 | 507 | } |
| 502 | 508 | |
| — | — | @@ -508,7 +514,9 @@ |
| 509 | 515 | |
| 510 | 516 | p = weightCombiner.apply(p, w); |
| 511 | 517 | |
| 512 | | - pop += p; //XXX: keep raw and processed pop |
| | 518 | + pop += p; |
| | 519 | + popf += popularityNormalizer.apply(p); |
| | 520 | + |
| 513 | 521 | weight += w; |
| 514 | 522 | } |
| 515 | 523 | |
| — | — | @@ -523,14 +531,15 @@ |
| 524 | 532 | |
| 525 | 533 | sim = n == 0 ? 0 : sim / n; //scale |
| 526 | 534 | pop = c == 0 ? 0 : pop / c; //scale |
| | 535 | + |
| | 536 | + simf = n == 0 ? 0 : simf / n; //scale |
| | 537 | + popf = c == 0 ? 0 : popf / c; //scale |
| | 538 | + |
| 527 | 539 | weight = c == 0 ? 0 : weight / c; //scale |
| 528 | 540 | |
| 529 | 541 | pop = doubleSanity(pop, "normal popularity", "check popularityMeasure!", 0, 0.1, Double.MAX_VALUE, 0); |
| 530 | 542 | sim = doubleSanity(sim, "normal average simility", "ooops!", 0, 0.1, 1, 0.1); |
| 531 | 543 | |
| 532 | | - double popf = popularityNormalizer.apply(pop); |
| 533 | | - double simf = similarityNormalizer.apply(sim); |
| 534 | | - |
| 535 | 544 | popf = doubleSanity(popf, "normal popularity", "check popularityNormalizer!", 0, 0.1, 1, 0.1); |
| 536 | 545 | simf = doubleSanity(simf, "normal similarity", "check similarityNormalizer!", 0, 0.1, 1, 0.1); |
| 537 | 546 | |