| Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/disambig/CoherenceDisambiguator.java |
| — | — | @@ -238,8 +238,6 @@ |
| 239 | 239 | return getScore(r.getInterpretation(), context, similarities, features); |
| 240 | 240 | } |
| 241 | 241 | |
| 242 | | - pruneMeanings(meanings); |
| 243 | | - |
| 244 | 242 | sz = meanings.size(); |
| 245 | 243 | if (context!=null) sz += context.size(); |
| 246 | 244 | if (sz <2) { |
| — | — | @@ -267,26 +265,34 @@ |
| 268 | 266 | return getBestInterpretation(root, meanings, context, interpretations, similarities, features); |
| 269 | 267 | } |
| 270 | 268 | |
| | 269 | + protected <X extends T>Map<X, List<? extends C>> getMeanings(Collection<X> terms) throws PersistenceException { |
| | 270 | + Map<X, List<? extends C>> meanings = super.getMeanings(terms); |
| | 271 | + pruneMeanings(meanings); |
| | 272 | + return meanings; |
| | 273 | + } |
| | 274 | + |
| 271 | 275 | protected void pruneMeanings(Map<? extends T, List<? extends C>> meanings) { |
| 272 | | - if (minPopularity<=1) return; //nothing to do |
| 273 | | - |
| 274 | 276 | Iterator<?> eit = meanings.entrySet().iterator(); |
| 275 | 277 | while (eit.hasNext()) { |
| 276 | 278 | Entry<T, List<? extends C>> e = (Entry<T, List<? extends C>>) eit.next(); //XXX: ugly cast. got confused about generics. ugh. |
| 277 | 279 | List<? extends C> m = e.getValue(); |
| 278 | 280 | if (m==null) continue; |
| 279 | 281 | |
| 280 | | - Iterator<? extends C> cit = m.iterator(); |
| 281 | | - while (cit.hasNext()) { |
| 282 | | - C c = cit.next(); |
| 283 | | - double p = popularityMeasure.measure(c); |
| 284 | | - |
| 285 | | - if (p<minPopularity) { |
| 286 | | - if (m.size()==1) { |
| 287 | | - eit.remove(); |
| 288 | | - break; |
| 289 | | - } else { |
| 290 | | - cit.remove(); |
| | 282 | + if (minPopularity>0) { |
| | 283 | + Iterator<? extends C> cit = m.iterator(); |
| | 284 | + while (cit.hasNext()) { |
| | 285 | + C c = cit.next(); |
| | 286 | + double p = popularityMeasure.measure(c); |
| | 287 | + |
| | 288 | + if (p<minPopularity) { |
| | 289 | + trace("pruning unpopular meaning of "+e.getKey()+" (pop: "+p+" < "+minPopularity+"): "+c.getName()); |
| | 290 | + |
| | 291 | + if (m.size()==1) { |
| | 292 | + eit.remove(); |
| | 293 | + break; |
| | 294 | + } else { |
| | 295 | + cit.remove(); |
| | 296 | + } |
| 291 | 297 | } |
| 292 | 298 | } |
| 293 | 299 | } |
| — | — | @@ -294,6 +300,9 @@ |
| 295 | 301 | if (m.size()==0) eit.remove(); |
| 296 | 302 | else if (m.size()>maxMeanings) { |
| 297 | 303 | Collections.sort(m, popularityComparator); |
| | 304 | + |
| | 305 | + trace("pruning least popular meanings of "+e.getKey()+" (keeping top "+maxMeanings+"): "+m.subList(maxMeanings, m.size())); |
| | 306 | + |
| 298 | 307 | m = m.subList(0, maxMeanings); |
| 299 | 308 | e.setValue(m); |
| 300 | 309 | } |
| — | — | @@ -321,6 +330,7 @@ |
| 322 | 331 | for (Disambiguator.Interpretation<X, C> interp: interpretations) { |
| 323 | 332 | CoherenceDisambiguation<X, C> r = getScore(interp, context, similarities, features); |
| 324 | 333 | double score = r.getScore(); |
| | 334 | + //trace(" ~ score "+score+": "+r.getMeanings()); |
| 325 | 335 | |
| 326 | 336 | if ( ( best == null && score> 0 && !Double.isNaN(score)) |
| 327 | 337 | || (score > bestScore && !Double.isNaN(score)) ) { |
| — | — | @@ -398,16 +408,16 @@ |
| 399 | 409 | |
| 400 | 410 | for (TermReference t: interp.getSequence()) { |
| 401 | 411 | C m = interp.getMeanings().get(t); |
| 402 | | - ((HashMap<TermReference, C>)concepts).put(t, m); |
| | 412 | + ((Map<TermReference, C>)concepts).put(t, m); |
| 403 | 413 | } |
| 404 | 414 | |
| 405 | 415 | if (context != null) { |
| 406 | 416 | for (C con: context) { |
| 407 | | - if (con!=null)((HashMap<TermReference, C>)concepts).put(new Term(con.getName(), 1), con); |
| | 417 | + if (con!=null)((Map<TermReference, C>)concepts).put(new Term(con.getName(), 1), con); |
| 408 | 418 | } |
| 409 | 419 | } |
| 410 | 420 | } else { |
| 411 | | - concepts = (HashMap<TermReference, C>)interp.getMeanings(); |
| | 421 | + concepts = (Map<TermReference, C>)interp.getMeanings(); |
| 412 | 422 | } |
| 413 | 423 | |
| 414 | 424 | int c = concepts.size(); |
| Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/disambig/SlidingCoherenceDisambiguator.java |
| — | — | @@ -79,8 +79,6 @@ |
| 80 | 80 | return getScore(r.getInterpretation(), context, similarities, features); |
| 81 | 81 | } |
| 82 | 82 | |
| 83 | | - pruneMeanings(meanings); |
| 84 | | - |
| 85 | 83 | sz = meanings.size(); |
| 86 | 84 | if (context!=null) sz += context.size(); |
| 87 | 85 | |