Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[4040] trunk/components-core/src/main/java/org/ dllearner/algorithms/isle
|
From: <lor...@us...> - 2013-09-03 16:48:19
|
Revision: 4040
http://sourceforge.net/p/dl-learner/code/4040
Author: lorenz_b
Date: 2013-09-03 16:48:16 +0000 (Tue, 03 Sep 2013)
Log Message:
-----------
Added first implementation of semantic index.
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SemanticAnnotator.java
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
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/PMIRelevanceMetric.java
Removed Paths:
-------------
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndexFactory.java
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SemanticAnnotator.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SemanticAnnotator.java 2013-09-03 16:09:27 UTC (rev 4039)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SemanticAnnotator.java 2013-09-03 16:48:16 UTC (rev 4040)
@@ -6,7 +6,6 @@
import org.dllearner.algorithms.isle.EntityCandidateGenerator;
import org.dllearner.algorithms.isle.WordSenseDisambiguation;
import org.dllearner.core.owl.Entity;
-import org.semanticweb.owlapi.model.OWLOntology;
/**
* Provides methods to annotate documents.
@@ -15,7 +14,6 @@
*/
public class SemanticAnnotator {
- private OWLOntology ontology;
private WordSenseDisambiguation wordSenseDisambiguation;
private EntityCandidateGenerator entityCandidateGenerator;
private LinguisticAnnotator linguisticAnnotator;
@@ -26,9 +24,8 @@
*
* @param ontology the ontology to use entities from
*/
- public SemanticAnnotator(OWLOntology ontology, WordSenseDisambiguation wordSenseDisambiguation,
+ public SemanticAnnotator(WordSenseDisambiguation wordSenseDisambiguation,
EntityCandidateGenerator entityCandidateGenerator, LinguisticAnnotator linguisticAnnotator) {
- this.ontology = ontology;
this.wordSenseDisambiguation = wordSenseDisambiguation;
this.entityCandidateGenerator = entityCandidateGenerator;
this.linguisticAnnotator = linguisticAnnotator;
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 2013-09-03 16:09:27 UTC (rev 4039)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/SemanticIndex.java 2013-09-03 16:48:16 UTC (rev 4040)
@@ -1,11 +1,19 @@
package org.dllearner.algorithms.isle.index.semantic;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Set;
+
+import org.dllearner.algorithms.isle.EntityCandidateGenerator;
+import org.dllearner.algorithms.isle.WordSenseDisambiguation;
import org.dllearner.algorithms.isle.index.AnnotatedDocument;
-import org.dllearner.algorithms.isle.index.Document;
+import org.dllearner.algorithms.isle.index.LinguisticAnnotator;
+import org.dllearner.algorithms.isle.index.SemanticAnnotator;
+import org.dllearner.algorithms.isle.index.TextDocument;
+import org.dllearner.algorithms.isle.index.syntactic.SyntacticIndex;
import org.dllearner.core.owl.Entity;
+import org.semanticweb.owlapi.model.OWLOntology;
-import java.util.Set;
-
/**
* Interface for an index which is able to resolve a given entity's URI to the set of documents containing
* this entity, i.e., documents which contain words disambiguated to the given entity.
@@ -13,14 +21,51 @@
* @author Lorenz Buehmann
* @author Daniel Fleischhacker
*/
-public interface SemanticIndex {
+public abstract class SemanticIndex {
+
+ 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;
+ }
+
+ /**
+ * Precompute the whole index, i.e. iterate over all entities and compute all annotated documents.
+ */
+ public void buildIndex(Set<TextDocument> documents){
+ for (TextDocument document : documents) {
+ 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);
+ }
+ }
+ }
+
/**
* 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);
+ public Set<AnnotatedDocument> getDocuments(Entity entity){
+ Set<AnnotatedDocument> annotatedDocuments = index.get(entity);
+ return annotatedDocuments;
+ }
/**
* Returns the number of documents for the given entity.
@@ -28,12 +73,16 @@
* @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);
+ public int count(Entity entity){
+ return index.get(entity).size();
+ }
/**
* Returns the total number of documents contained in the index.
*
* @return the total number of documents contained in the index
*/
- public int getSize();
+ 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 2013-09-03 16:09:27 UTC (rev 4039)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndex.java 2013-09-03 16:48:16 UTC (rev 4040)
@@ -3,28 +3,20 @@
*/
package org.dllearner.algorithms.isle.index.semantic.simple;
-import org.dllearner.algorithms.isle.index.AnnotatedDocument;
-import org.dllearner.algorithms.isle.index.Document;
+import org.dllearner.algorithms.isle.RandomWordSenseDisambiguation;
+import org.dllearner.algorithms.isle.index.SimpleEntityCandidateGenerator;
+import org.dllearner.algorithms.isle.index.SimpleLinguisticAnnotator;
+import org.dllearner.algorithms.isle.index.semantic.SemanticIndex;
import org.dllearner.algorithms.isle.index.syntactic.SyntacticIndex;
-import org.dllearner.algorithms.isle.index.semantic.SemanticIndex;
-import org.dllearner.algorithms.isle.textretrieval.RDFSLabelEntityTextRetriever;
-import org.dllearner.core.owl.Entity;
import org.semanticweb.owlapi.model.OWLOntology;
-import java.util.HashSet;
-import java.util.Map;
-import java.util.Map.Entry;
-import java.util.Set;
-
/**
* A semantic index which returns all documents which contain at least one of the labels assigned to a specific
* entity in a provided ontology.
*
* @author Lorenz Buehmann
*/
-public class SimpleSemanticIndex implements SemanticIndex {
- private SyntacticIndex syntacticIndex;
- private RDFSLabelEntityTextRetriever labelRetriever;
+public class SimpleSemanticIndex extends SemanticIndex {
/**
* Initializes the semantic index to use {@code ontology} for finding all labels of an entity and
@@ -34,41 +26,11 @@
* @param syntacticIndex index to query for documents containing the labels
*/
public SimpleSemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex) {
- this.syntacticIndex = syntacticIndex;
- labelRetriever = new RDFSLabelEntityTextRetriever(ontology);
+ super(ontology,
+ syntacticIndex,
+ new RandomWordSenseDisambiguation(ontology),
+ new SimpleEntityCandidateGenerator(ontology),
+ new SimpleLinguisticAnnotator());
}
- /* (non-Javadoc)
- * @see org.dllearner.algorithms.isle.SemanticIndex#getDocuments(org.dllearner.core.owl.Entity)
- */
- @Override
- public Set<AnnotatedDocument> getDocuments(Entity entity) {
- Set<AnnotatedDocument> documents = new HashSet<AnnotatedDocument>();
- Map<String, Double> relevantText = labelRetriever.getRelevantText(entity);
-
- for (Entry<String, Double> entry : relevantText.entrySet()) {
- String label = entry.getKey();
- documents.addAll(syntacticIndex.getDocuments(label));
- }
-
- return documents;
- }
-
- /* (non-Javadoc)
- * @see org.dllearner.algorithms.isle.SemanticIndex#count(java.lang.String)
- */
- @Override
- public int count(Entity entity) {
- return getDocuments(entity).size();
- }
-
- /* (non-Javadoc)
- * @see org.dllearner.algorithms.isle.SemanticIndex#getSize()
- */
- @Override
- public int getSize() {
- return syntacticIndex.getSize();
- }
-
-
}
Deleted: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndexFactory.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndexFactory.java 2013-09-03 16:09:27 UTC (rev 4039)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/semantic/simple/SimpleSemanticIndexFactory.java 2013-09-03 16:48:16 UTC (rev 4040)
@@ -1,40 +0,0 @@
-/**
- *
- */
-package org.dllearner.algorithms.isle.index.semantic.simple;
-
-import org.dllearner.algorithms.isle.index.syntactic.SyntacticIndex;
-import org.dllearner.algorithms.isle.index.semantic.SemanticIndex;
-import org.dllearner.algorithms.isle.index.semantic.SemanticIndexFactory;
-import org.semanticweb.owlapi.model.OWLOntology;
-
-import java.io.File;
-
-/**
- * This gets a syntactic index and returns a semantic index by applying WSD etc.
- *
- * @author Lorenz Buehmann
- * @author Daniel Fleischhacker
- */
-public class SimpleSemanticIndexFactory implements SemanticIndexFactory {
- private OWLOntology ontology;
- private SyntacticIndex syntacticIndex;
-
- /**
- * Initializes a semantic index factory for creating simple semantic indexes. Simple semantic indexes use
- * the labels assigned to an entity in {@code ontology} as its surface forms and return the all documents
- * from the given syntactic index which contain at least one of these surface forms.
- *
- * @param syntacticIndex the syntactic index in which occurrences of the labels are searched
- * @param ontology the ontology retrieve the entities' labels from
- */
- public SimpleSemanticIndexFactory(SyntacticIndex syntacticIndex, OWLOntology ontology) {
- this.syntacticIndex = syntacticIndex;
- this.ontology = ontology;
- }
-
- @Override
- public SemanticIndex createIndex(File inputDirectory) {
- return new SimpleSemanticIndex(ontology, syntacticIndex);
- }
-}
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/PMIRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/PMIRelevanceMetric.java 2013-09-03 16:09:27 UTC (rev 4039)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/PMIRelevanceMetric.java 2013-09-03 16:48:16 UTC (rev 4040)
@@ -3,12 +3,13 @@
*/
package org.dllearner.algorithms.isle.metrics;
-import com.google.common.collect.Sets;
-import org.dllearner.algorithms.isle.index.Document;
+import java.util.Set;
+
+import org.dllearner.algorithms.isle.index.AnnotatedDocument;
import org.dllearner.algorithms.isle.index.semantic.SemanticIndex;
import org.dllearner.core.owl.Entity;
-import java.util.Set;
+import com.google.common.collect.Sets;
/**
* @author Lorenz Buehmann
@@ -22,9 +23,9 @@
@Override
public double getRelevance(Entity entityA, Entity entityB){
- Set<Document> documentsA = index.getDocuments(entityA);
- Set<Document> documentsB = index.getDocuments(entityB);
- Set<Document> documentsAB = Sets.intersection(documentsA, documentsB);
+ Set<AnnotatedDocument> documentsA = index.getDocuments(entityA);
+ Set<AnnotatedDocument> documentsB = index.getDocuments(entityB);
+ Set<AnnotatedDocument> documentsAB = Sets.intersection(documentsA, documentsB);
int nrOfDocuments = index.getSize();
double dPClass = nrOfDocuments == 0 ? 0 : ((double) documentsA.size() / (double) nrOfDocuments);
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
