From: <dfl...@us...> - 2013-10-29 14:11:10
|
Revision: 4133 http://sourceforge.net/p/dl-learner/code/4133 Author: dfleischhacker Date: 2013-10-29 14:11:07 +0000 (Tue, 29 Oct 2013) Log Message: ----------- Bug searching Modified Paths: -------------- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/Annotation.java trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieEntityCandidateGenerator.java trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieLinguisticAnnotator.java trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/Annotation.java =================================================================== --- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/Annotation.java 2013-10-29 13:23:45 UTC (rev 4132) +++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/Annotation.java 2013-10-29 14:11:07 UTC (rev 4133) @@ -14,8 +14,17 @@ private Document referencedDocument; private int offset; private int length; - - public Annotation(Document referencedDocument, int offset, int length) { + private String matchedString; + + public String getMatchedString() { + return matchedString; + } + + public void setMatchedString(String matchedString) { + this.matchedString = matchedString; + } + + public Annotation(Document referencedDocument, int offset, int length) { this.referencedDocument = referencedDocument; this.offset = offset; this.length = length; Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieEntityCandidateGenerator.java =================================================================== --- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieEntityCandidateGenerator.java 2013-10-29 13:23:45 UTC (rev 4132) +++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieEntityCandidateGenerator.java 2013-10-29 14:11:07 UTC (rev 4133) @@ -1,20 +1,13 @@ package org.dllearner.algorithms.isle.index; -import java.util.ArrayList; -import java.util.Collections; -import java.util.Comparator; -import java.util.HashMap; -import java.util.HashSet; -import java.util.List; -import java.util.Set; -import java.util.regex.Pattern; - +import com.google.common.collect.Sets; import org.dllearner.algorithms.isle.EntityCandidateGenerator; import org.dllearner.algorithms.isle.StopWordFilter; import org.dllearner.core.owl.Entity; import org.semanticweb.owlapi.model.OWLOntology; -import com.google.common.collect.Sets; +import java.util.*; +import java.util.regex.Pattern; /** * Generates candidates using a entity candidates prefix trie @@ -33,7 +26,7 @@ } public Set<Entity> getCandidates(Annotation annotation) { - return candidatesTrie.getCandidateEntities(annotation.getToken()); + return candidatesTrie.getCandidateEntities(annotation.getMatchedString()); } /** @@ -131,7 +124,7 @@ for (Annotation annotation: annotations) candidatesMap.put(annotation, getCandidates(annotation)); - postProcess(candidatesMap, window, stopWordFilter); + //postProcess(candidatesMap, window, stopWordFilter); return candidatesMap; } Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieLinguisticAnnotator.java =================================================================== --- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieLinguisticAnnotator.java 2013-10-29 13:23:45 UTC (rev 4132) +++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/TrieLinguisticAnnotator.java 2013-10-29 14:11:07 UTC (rev 4133) @@ -36,6 +36,7 @@ String match = candidatesTrie.getLongestMatchingText(unparsed); if (match != null && !match.isEmpty()) { Annotation annotation = mapper.getOriginalAnnotationForPosition(i, match.length()); + annotation.setMatchedString(match); annotations.add(annotation); i += match.length() - 1; } Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java =================================================================== --- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java 2013-10-29 13:23:45 UTC (rev 4132) +++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java 2013-10-29 14:11:07 UTC (rev 4133) @@ -53,7 +53,7 @@ trie = new SimpleEntityCandidatesTrie(new RDFSLabelEntityTextRetriever(ontology), ontology, new SimpleEntityCandidatesTrie.DummyNameGenerator()); } -// trie.printTrie(); + trie.printTrie(); TrieLinguisticAnnotator linguisticAnnotator = new TrieLinguisticAnnotator(trie); linguisticAnnotator.setNormalizeWords(useWordNormalization); setSemanticAnnotator(new SemanticAnnotator( This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site. |