Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[4081] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle/index/semantic
|
From: <dfl...@us...> - 2013-09-05 07:46:37
|
Revision: 4081
http://sourceforge.net/p/dl-learner/code/4081
Author: dfleischhacker
Date: 2013-09-05 07:46:34 +0000 (Thu, 05 Sep 2013)
Log Message:
-----------
Pass ontology into SimpleSemanticIndex
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/SemanticIndex.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/semantic/SemanticIndex.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/SemanticIndex.java 2013-09-05 07:22:38 UTC (rev 4080)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/SemanticIndex.java 2013-09-05 07:46:34 UTC (rev 4081)
@@ -28,103 +28,101 @@
* @author Daniel Fleischhacker
*/
public abstract class SemanticIndex {
-
-
- private static final Logger logger = Logger.getLogger(SemanticIndex.class.getName());
-
- private SemanticAnnotator semanticAnnotator;
- private SyntacticIndex syntacticIndex;
- private Map<Entity, Set<AnnotatedDocument>> index;
- private OWLOntology ontology;
-
- public SemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex, WordSenseDisambiguation wordSenseDisambiguation,
- EntityCandidateGenerator entityCandidateGenerator, LinguisticAnnotator linguisticAnnotator) {
- this.ontology = ontology;
- this.syntacticIndex = syntacticIndex;
- semanticAnnotator = new SemanticAnnotator(wordSenseDisambiguation, entityCandidateGenerator, linguisticAnnotator);
- }
-
- public SemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex, SemanticAnnotator semanticAnnotator) {
- this.semanticAnnotator = semanticAnnotator;
- }
-
- public SemanticIndex() {
-}
-
- /**
- * @param semanticAnnotator the semanticAnnotator to set
- */
- public void setSemanticAnnotator(SemanticAnnotator semanticAnnotator) {
- this.semanticAnnotator = semanticAnnotator;
- }
-
- /**
- * Precompute the whole index, i.e. iterate over all entities and compute all annotated documents.
- */
- public void buildIndex(Set<TextDocument> documents){
- logger.info("Creating semantic index...");
- index = new HashMap<Entity, Set<AnnotatedDocument>>();
- for (TextDocument document : documents) {
- logger.info("Processing document:\n" + document);
- AnnotatedDocument annotatedDocument = semanticAnnotator.processDocument(document);
- for (Entity entity : annotatedDocument.getContainedEntities()) {
- Set<AnnotatedDocument> existingAnnotatedDocuments = index.get(entity);
- if(existingAnnotatedDocuments == null){
- existingAnnotatedDocuments = new HashSet<AnnotatedDocument>();
- index.put(entity, existingAnnotatedDocuments);
- }
- existingAnnotatedDocuments.add(annotatedDocument);
- }
- }
- logger.info("...done.");
- }
-
- public void buildIndex(OWLAnnotationProperty annotationProperty, String language){
- Set<OWLEntity> schemaEntities = new HashSet<OWLEntity>();
- schemaEntities.addAll(ontology.getClassesInSignature());
- schemaEntities.addAll(ontology.getObjectPropertiesInSignature());
- schemaEntities.addAll(ontology.getDataPropertiesInSignature());
- Set<TextDocument> documents = new HashSet<TextDocument>();
- for (OWLEntity entity : schemaEntities) {
- String label = null;
- Set<OWLAnnotation> annotations = entity.getAnnotations(ontology, annotationProperty);
- for (OWLAnnotation annotation : annotations) {
- if (annotation.getValue() instanceof OWLLiteral) {
- OWLLiteral val = (OWLLiteral) annotation.getValue();
- if (language != null) {
- if(val.hasLang(language)){
- label = val.getLiteral();
- }
-
- } else {
- label = val.getLiteral();
- }
- }
- }
- if(label != null){
- documents.add(new TextDocument(label));
- }
- }
- buildIndex(documents);
- }
-
+
+
+ private static final Logger logger = Logger.getLogger(SemanticIndex.class.getName());
+
+ private SemanticAnnotator semanticAnnotator;
+ private SyntacticIndex syntacticIndex;
+ private Map<Entity, Set<AnnotatedDocument>> index;
+ private OWLOntology ontology;
+
+ public SemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex, WordSenseDisambiguation wordSenseDisambiguation,
+ EntityCandidateGenerator entityCandidateGenerator, LinguisticAnnotator linguisticAnnotator) {
+ this.ontology = ontology;
+ this.syntacticIndex = syntacticIndex;
+ semanticAnnotator = new SemanticAnnotator(wordSenseDisambiguation, entityCandidateGenerator, linguisticAnnotator);
+ }
+
+ public SemanticIndex(OWLOntology ontology) {
+ this.ontology = ontology;
+ }
+
/**
+ * @param semanticAnnotator the semanticAnnotator to set
+ */
+ public void setSemanticAnnotator(SemanticAnnotator semanticAnnotator) {
+ this.semanticAnnotator = semanticAnnotator;
+ }
+
+ /**
+ * Precompute the whole index, i.e. iterate over all entities and compute all annotated documents.
+ */
+ public void buildIndex(Set<TextDocument> documents) {
+ logger.info("Creating semantic index...");
+ index = new HashMap<Entity, Set<AnnotatedDocument>>();
+ for (TextDocument document : documents) {
+ logger.info("Processing document:\n" + document);
+ AnnotatedDocument annotatedDocument = semanticAnnotator.processDocument(document);
+ for (Entity entity : annotatedDocument.getContainedEntities()) {
+ Set<AnnotatedDocument> existingAnnotatedDocuments = index.get(entity);
+ if (existingAnnotatedDocuments == null) {
+ existingAnnotatedDocuments = new HashSet<AnnotatedDocument>();
+ index.put(entity, existingAnnotatedDocuments);
+ }
+ existingAnnotatedDocuments.add(annotatedDocument);
+ }
+ }
+ logger.info("...done.");
+ }
+
+ public void buildIndex(OWLAnnotationProperty annotationProperty, String language) {
+ Set<OWLEntity> schemaEntities = new HashSet<OWLEntity>();
+ schemaEntities.addAll(ontology.getClassesInSignature());
+ schemaEntities.addAll(ontology.getObjectPropertiesInSignature());
+ schemaEntities.addAll(ontology.getDataPropertiesInSignature());
+ Set<TextDocument> documents = new HashSet<TextDocument>();
+ for (OWLEntity entity : schemaEntities) {
+ String label = null;
+ Set<OWLAnnotation> annotations = entity.getAnnotations(ontology, annotationProperty);
+ for (OWLAnnotation annotation : annotations) {
+ if (annotation.getValue() instanceof OWLLiteral) {
+ OWLLiteral val = (OWLLiteral) annotation.getValue();
+ if (language != null) {
+ if (val.hasLang(language)) {
+ label = val.getLiteral();
+ }
+
+ }
+ else {
+ label = val.getLiteral();
+ }
+ }
+ }
+ if (label != null) {
+ documents.add(new TextDocument(label));
+ }
+ }
+ buildIndex(documents);
+ }
+
+ /**
* Returns the set of annotated documents which reference the given entity using one of its surface forms.
*
* @param entity entity to retrieve documents
* @return documents referencing given entity
*/
- public Set<AnnotatedDocument> getDocuments(Entity entity){
- if(index == null){
- System.err.println("You have to prebuild the index before you can use this method.");
- System.exit(1);
- }
-
- Set<AnnotatedDocument> annotatedDocuments = index.get(entity);
- if(annotatedDocuments == null) {
- annotatedDocuments = new HashSet<AnnotatedDocument>();
- }
- return annotatedDocuments;
+ public Set<AnnotatedDocument> getDocuments(Entity entity) {
+ if (index == null) {
+ System.err.println("You have to prebuild the index before you can use this method.");
+ System.exit(1);
+ }
+
+ Set<AnnotatedDocument> annotatedDocuments = index.get(entity);
+ if (annotatedDocuments == null) {
+ annotatedDocuments = new HashSet<AnnotatedDocument>();
+ }
+ return annotatedDocuments;
}
/**
@@ -133,8 +131,8 @@
* @param entity entity to return number of referencing documents for
* @return number of documents for the given entity in this index
*/
- public int count(Entity entity){
- return index.get(entity).size();
+ public int count(Entity entity) {
+ return index.get(entity).size();
}
/**
@@ -142,7 +140,7 @@
*
* @return the total number of documents contained in the index
*/
- public int getSize(){
- return index.size();
+ public int getSize() {
+ return index.size();
}
}
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 2013-09-05 07:22:38 UTC (rev 4080)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java 2013-09-05 07:46:34 UTC (rev 4081)
@@ -29,8 +29,9 @@
* @param syntacticIndex index to query for documents containing the labels
*/
public SimpleSemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex) {
- SimpleEntityCandidatesTrie trie = new SimpleEntityCandidatesTrie(new RDFSLabelEntityTextRetriever(ontology), ontology);
- setSemanticAnnotator(new SemanticAnnotator(
+ super(ontology);
+ SimpleEntityCandidatesTrie trie = new SimpleEntityCandidatesTrie(new RDFSLabelEntityTextRetriever(ontology), ontology);
+ setSemanticAnnotator(new SemanticAnnotator(
new SimpleWordSenseDisambiguation(ontology),
new TrieEntityCandidateGenerator(ontology, trie),
new TrieLinguisticAnnotator(trie)));
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
