Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[4221] trunk/components-core/src
|
From: <and...@us...> - 2014-02-11 16:48:05
|
Revision: 4221
http://sourceforge.net/p/dl-learner/code/4221
Author: andremelo
Date: 2014-02-11 16:48:02 +0000 (Tue, 11 Feb 2014)
Log Message:
-----------
Adding relevance measures from http://www.cse.iitb.ac.in/~damani/papers/conll13.pdf
Added Paths:
-----------
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/ChiSquareRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/DiceRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/JaccardRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/LLRRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SCIRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SignificantPMIRelevanceMetric.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/TTestRelevanceMetric.java
trunk/components-core/src/test/java/org/dllearner/algorithms/isle/metrics/RelevanceMetricsTest.java
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/ChiSquareRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/ChiSquareRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/ChiSquareRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,52 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * Chi Squared
+ * @author Andre Melo
+ *
+ */
+public class ChiSquareRelevanceMetric extends AbstractRelevanceMetric {
+
+ public ChiSquareRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ private double chiSquareIteration(double fXY, double e_fXY) {
+ return Math.pow(fXY - e_fXY,2)/e_fXY;
+ }
+
+ @Override
+ public synchronized double getRelevance(Entity entityA, Entity entityB){
+ double fA = index.getNumberOfDocumentsFor(entityA);
+ double fB = index.getNumberOfDocumentsFor(entityB);
+ double fAB = index.getNumberOfDocumentsFor(entityA, entityB);
+ double N = index.getTotalNumberOfDocuments();
+
+ double e_fAB = fA*fB/N; // Expected frequency of A and B assuming independence
+
+ double chi2 = 0;
+
+ // X=A and Y=B
+ chi2 += chiSquareIteration(fAB, e_fAB);
+ // X=A and Y=not B
+ chi2 += chiSquareIteration(fA-fAB, fA-e_fAB);
+ // X=not A and Y=B
+ chi2 += chiSquareIteration(fB-fAB, fB-e_fAB);
+ // X=not A and Y=not B
+ chi2 += chiSquareIteration(N-fA-fB+fAB, N-fA-fB+e_fAB);
+
+ return chi2;
+ }
+
+ @Override
+ public synchronized double getNormalizedRelevance(Entity entityA, Entity entityB){
+ return Double.NaN;
+ }
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/DiceRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/DiceRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/DiceRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,33 @@
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * @author Andre Melo
+ *
+ */
+public class DiceRelevanceMetric extends AbstractRelevanceMetric{
+
+ public DiceRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ @Override
+ public double getRelevance(Entity entityA, Entity entityB) {
+ double nrOfDocumentsA = index.getNumberOfDocumentsFor(entityA);
+ double nrOfDocumentsB = index.getNumberOfDocumentsFor(entityB);
+ double nrOfDocumentsAandB = index.getNumberOfDocumentsFor(entityA, entityB);
+
+ double dice = 2 * nrOfDocumentsAandB / (nrOfDocumentsA + nrOfDocumentsB);
+
+ return dice;
+ }
+
+ @Override
+ public double getNormalizedRelevance(Entity entity1, Entity entity2) {
+ return getRelevance(entity1, entity2);
+ }
+
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/JaccardRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/JaccardRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/JaccardRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,37 @@
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * @author Andre Melo
+ *
+ */
+public class JaccardRelevanceMetric extends AbstractRelevanceMetric{
+
+ public JaccardRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ @Override
+ public double getRelevance(Entity entityA, Entity entityB) {
+ long nrOfDocumentsA = index.getNumberOfDocumentsFor(entityA);
+ long nrOfDocumentsB = index.getNumberOfDocumentsFor(entityB);
+ double nrOfDocumentsAandB = index.getNumberOfDocumentsFor(entityA, entityB);
+ double nrOfDocumentsAorB = nrOfDocumentsA + nrOfDocumentsB - nrOfDocumentsAandB;
+
+ if (nrOfDocumentsAorB==0)
+ return 1.0;
+
+ double jaccard = nrOfDocumentsAandB / nrOfDocumentsAorB;
+
+ return jaccard;
+ }
+
+ @Override
+ public double getNormalizedRelevance(Entity entity1, Entity entity2) {
+ return getRelevance(entity1, entity2);
+ }
+
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/LLRRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/LLRRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/LLRRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,54 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * Log Likelihood Ratio
+ * @author Andre Melo
+ *
+ */
+public class LLRRelevanceMetric extends AbstractRelevanceMetric {
+
+ public LLRRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ private double llrIteration(double pXY, double pX, double pY) {
+ return pXY * Math.log(pXY/(pX*pY));
+ }
+
+ @Override
+ public synchronized double getRelevance(Entity entityA, Entity entityB){
+ double fA = index.getNumberOfDocumentsFor(entityA);
+ double fB = index.getNumberOfDocumentsFor(entityB);
+ double fAB = index.getNumberOfDocumentsFor(entityA, entityB);
+ double N = index.getTotalNumberOfDocuments();
+
+ double pA = fA/N;
+ double pB = fB/N;
+ double pAB = fAB/N;
+
+ double llr = 0;
+
+ // X=A and Y=B
+ llr += llrIteration( pAB, pA, pB );
+ // X=A and Y=not B
+ llr += llrIteration( pA-pAB, pA, 1-pB );
+ // X=not A and Y=B
+ llr += llrIteration( pB-pAB, 1-pA, pB );
+ // X=not A and Y=not B
+ llr += llrIteration( 1-pA-pB+pAB,1-pA, 1-pB );
+
+ return llr;
+ }
+
+ @Override
+ public synchronized double getNormalizedRelevance(Entity entityA, Entity entityB){
+ return Double.NaN;
+ }
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SCIRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SCIRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SCIRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,42 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * @author Andre Melo
+ *
+ */
+public class SCIRelevanceMetric extends AbstractRelevanceMetric {
+
+ public SCIRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ @Override
+ public synchronized double getRelevance(Entity entityA, Entity entityB){
+ long nrOfDocumentsA = index.getNumberOfDocumentsFor(entityA);
+ long nrOfDocumentsB = index.getNumberOfDocumentsFor(entityB);
+ long nrOfDocumentsAB = index.getNumberOfDocumentsFor(entityA, entityB);
+
+ long nrOfDocuments = index.getTotalNumberOfDocuments();
+
+ double pA = nrOfDocuments == 0 ? 0 : ((double) nrOfDocumentsA / (double) nrOfDocuments);
+ double pB = nrOfDocuments == 0 ? 0 : ((double) nrOfDocumentsB / (double) nrOfDocuments);
+ double pAB = nrOfDocuments == 0 ? 0 : ((double) nrOfDocumentsAB / (double) nrOfDocuments);
+
+ double sci = pAB / (pA * Math.sqrt(pB));
+
+ return sci;
+ }
+
+ @Override
+ public synchronized double getNormalizedRelevance(Entity entityA, Entity entityB){
+ //TODO
+ return Double.NaN;
+ }
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SignificantPMIRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SignificantPMIRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/SignificantPMIRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,48 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * @author Lorenz Buehmann
+ *
+ */
+public class SignificantPMIRelevanceMetric extends AbstractRelevanceMetric {
+
+ protected final double delta;
+
+ /**
+ *
+ * @param index: semantic index
+ * @param delta: parameter varying from 0 to 1
+ */
+ public SignificantPMIRelevanceMetric(Index index, double delta) {
+ super(index);
+ if (delta<0 ||delta>1)
+ throw new IllegalArgumentException("Delta parameter should be in [0,1]");
+ this.delta = delta;
+ }
+
+ @Override
+ public synchronized double getRelevance(Entity entityA, Entity entityB){
+ double fA = index.getNumberOfDocumentsFor(entityA);
+ double fB = index.getNumberOfDocumentsFor(entityB);
+ double fAB = index.getNumberOfDocumentsFor(entityA, entityB);
+
+ double N = index.getTotalNumberOfDocuments();
+
+ double pmi = Math.log(fAB / (fA*fB/N + Math.sqrt(fA)*Math.sqrt(Math.log(delta)/-2)));
+
+ return pmi;
+ }
+
+ @Override
+ public synchronized double getNormalizedRelevance(Entity entityA, Entity entityB){
+ //TODO
+ return Double.NaN;
+ }
+
+}
Added: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/TTestRelevanceMetric.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/TTestRelevanceMetric.java (rev 0)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/metrics/TTestRelevanceMetric.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,39 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.core.owl.Entity;
+
+/**
+ * @author Andre Melo
+ *
+ */
+public class TTestRelevanceMetric extends AbstractRelevanceMetric {
+
+ public TTestRelevanceMetric(Index index) {
+ super(index);
+ }
+
+ @Override
+ public synchronized double getRelevance(Entity entityA, Entity entityB){
+ double nrOfDocumentsA = index.getNumberOfDocumentsFor(entityA);
+ double nrOfDocumentsB = index.getNumberOfDocumentsFor(entityB);
+ double nrOfDocumentsAB = index.getNumberOfDocumentsFor(entityA, entityB);
+
+ double nrOfDocuments = index.getTotalNumberOfDocuments();
+
+ double ttest = (nrOfDocumentsAB - (nrOfDocumentsA*nrOfDocumentsB)/nrOfDocuments) /
+ Math.sqrt(nrOfDocumentsAB*(1-nrOfDocumentsAB/nrOfDocuments));
+
+ return ttest;
+ }
+
+ @Override
+ public synchronized double getNormalizedRelevance(Entity entityA, Entity entityB){
+ //TODO
+ return Double.NaN;
+ }
+
+}
Added: trunk/components-core/src/test/java/org/dllearner/algorithms/isle/metrics/RelevanceMetricsTest.java
===================================================================
--- trunk/components-core/src/test/java/org/dllearner/algorithms/isle/metrics/RelevanceMetricsTest.java (rev 0)
+++ trunk/components-core/src/test/java/org/dllearner/algorithms/isle/metrics/RelevanceMetricsTest.java 2014-02-11 16:48:02 UTC (rev 4221)
@@ -0,0 +1,140 @@
+/**
+ *
+ */
+package org.dllearner.algorithms.isle.metrics;
+
+import static org.junit.Assert.fail;
+
+import java.io.BufferedInputStream;
+import java.io.InputStream;
+import java.net.URL;
+
+import org.apache.commons.compress.compressors.CompressorInputStream;
+import org.apache.commons.compress.compressors.CompressorStreamFactory;
+import org.dllearner.algorithms.isle.index.Index;
+import org.dllearner.algorithms.isle.index.syntactic.SolrSyntacticIndex;
+import org.dllearner.core.owl.Entity;
+import org.dllearner.core.owl.NamedClass;
+import org.dllearner.core.owl.ObjectProperty;
+import org.junit.Test;
+import org.semanticweb.owlapi.apibinding.OWLManager;
+import org.semanticweb.owlapi.model.OWLOntology;
+
+/**
+ * @author Lorenz Buehmann
+ *
+ */
+public class RelevanceMetricsTest {
+
+ AbstractRelevanceMetric metric;
+ Index index;
+ static final String solrServerURL = "http://solr.aksw.org/en_dbpedia_resources/";
+ static final String searchField = "comment";
+ static final String DBPEDIA_NS = "http://dbpedia.org/ontology/";
+
+ /**
+ *
+ */
+ public RelevanceMetricsTest() {
+ OWLOntology ontology = null;
+ try {
+ URL url = new URL("http://downloads.dbpedia.org/3.9/dbpedia_3.9.owl.bz2");
+ InputStream is = new BufferedInputStream(url.openStream());
+ CompressorInputStream in = new CompressorStreamFactory().createCompressorInputStream("bzip2", is);
+ ontology = OWLManager.createOWLOntologyManager().loadOntologyFromOntologyDocument(in);
+ } catch (Exception e){
+ e.printStackTrace();
+ }
+ index = new SolrSyntacticIndex(ontology, solrServerURL, searchField);
+
+ }
+
+ private void computeRelevanceScores(AbstractRelevanceMetric metric) {
+ //dbo:Person and dbo:Film
+ Entity entity1 = new NamedClass(DBPEDIA_NS + "Person");
+ Entity entity2 = new NamedClass(DBPEDIA_NS + "Film");
+ double relevance = metric.getRelevance(entity1, entity2);
+ System.out.println(relevance);
+
+ //dbo:Person and dbo:Animal
+ entity1 = new NamedClass(DBPEDIA_NS + "Person");
+ entity2 = new NamedClass(DBPEDIA_NS + "Animal");
+ relevance = metric.getRelevance(entity1, entity2);
+ System.out.println(relevance);
+
+ // dbo:Person and dbo:Animal
+ entity1 = new NamedClass(DBPEDIA_NS + "Person");
+ entity2 = new ObjectProperty(DBPEDIA_NS + "birthPlace");
+ relevance = metric.getRelevance(entity1, entity2);
+ System.out.println(relevance);
+ }
+
+ /**
+ * Test method for {@link org.dllearner.algorithms.isle.metrics.PMIRelevanceMetric#getRelevance(org.dllearner.core.owl.Entity, org.dllearner.core.owl.Entity)}.
+ */
+ @Test
+ public void testGetRelevanceJaccard() {
+ System.out.println("JACCARD: ");
+ metric = new JaccardRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevancePMI() {
+ System.out.println("PMI: ");
+ metric = new PMIRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceSignificantPMI() {
+ System.out.println("SignificantPMI: ");
+ double delta = 0.5;
+ metric = new SignificantPMIRelevanceMetric(index,delta);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceDice() {
+ System.out.println("DICE: ");
+ metric = new DiceRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceSCI() {
+ System.out.println("SCI: ");
+ metric = new SCIRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceTTest() {
+ System.out.println("T-TEST: ");
+ metric = new TTestRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceChiSquared() {
+ System.out.println("CHI^2: ");
+ metric = new ChiSquareRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ @Test
+ public void testGetRelevanceLLR() {
+ System.out.println("LLR: ");
+ metric = new LLRRelevanceMetric(index);
+ computeRelevanceScores(metric);
+ }
+
+ /**
+ * Test method for {@link org.dllearner.algorithms.isle.metrics.PMIRelevanceMetric#getNormalizedRelevance(org.dllearner.core.owl.Entity, org.dllearner.core.owl.Entity)}.
+ */
+ @Test
+ public void testGetNormalizedRelevance() {
+ fail("Not yet implemented");
+ }
+
+}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
