Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[2635] trunk/components-core/src
|
From: <jen...@us...> - 2011-01-31 19:37:43
|
Revision: 2635
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=2635&view=rev
Author: jenslehmann
Date: 2011-01-31 19:37:37 +0000 (Mon, 31 Jan 2011)
Log Message:
-----------
started unit tests for pos-neg learning measures
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/learningproblems/PosNegLPStandard.java
trunk/components-core/src/test/java/org/dllearner/test/junit/HeuristicTests.java
Modified: trunk/components-core/src/main/java/org/dllearner/learningproblems/PosNegLPStandard.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/learningproblems/PosNegLPStandard.java 2011-01-26 18:39:46 UTC (rev 2634)
+++ trunk/components-core/src/main/java/org/dllearner/learningproblems/PosNegLPStandard.java 2011-01-31 19:37:37 UTC (rev 2635)
@@ -313,18 +313,7 @@
@Override
public double getAccuracyOrTooWeak(Description description, double noise) {
// delegates to the appropriate methods
- return useApproximations ? getAccuracyOrTooWeakApprox(description, noise) : getAccuracyOrTooWeakExact(description, noise);
- /*
- if(useApproximations) {
- if(useFMeasure) {
- return getFMeasureOrTooWeakApprox(description, noise);
- } else {
- throw new Error("approximating pred. acc not implemented");
- }
- } else {
- return getPredAccuracyOrTooWeakExact(description, noise);
- }
- */
+ return useApproximations ? getAccuracyOrTooWeakApprox(description, noise) : getAccuracyOrTooWeakExact(description, noise);
}
public double getAccuracyOrTooWeakApprox(Description description, double noise) {
@@ -386,6 +375,8 @@
return ret;
} else if(heuristic.equals(HeuristicType.FMEASURE)) {
+ System.out.println("Testing " + description);
+
// we abort when there are too many uncovered positives
int maxNotCovered = (int) Math.ceil(noise*positiveExamples.size());
int instancesCovered = 0;
@@ -437,6 +428,8 @@
if(heuristic.equals(HeuristicType.PRED_ACC)) {
return getPredAccuracyOrTooWeakExact(description, noise);
} else if(heuristic.equals(HeuristicType.FMEASURE)) {
+ return getFMeasureOrTooWeakExact(description, noise);
+ /*
// computing R(C) restricted to relevant instances
int additionalInstances = 0;
for(Individual ind : negativeExamples) {
@@ -457,6 +450,7 @@
double precision = (additionalInstances + coveredInstances == 0) ? 0 : coveredInstances / (double) (coveredInstances + additionalInstances);
return Heuristics.getFScore(recall, precision);
+ */
} else {
throw new Error("Heuristic " + heuristic + " not implemented.");
}
@@ -518,7 +512,8 @@
double precision = (additionalInstances + coveredInstances == 0) ? 0 : coveredInstances / (double) (coveredInstances + additionalInstances);
- return getFMeasure(recall, precision);
+// return getFMeasure(recall, precision);
+ return Heuristics.getFScore(recall, precision);
}
// instead of using the standard operation, we use optimisation
Modified: trunk/components-core/src/test/java/org/dllearner/test/junit/HeuristicTests.java
===================================================================
--- trunk/components-core/src/test/java/org/dllearner/test/junit/HeuristicTests.java 2011-01-26 18:39:46 UTC (rev 2634)
+++ trunk/components-core/src/test/java/org/dllearner/test/junit/HeuristicTests.java 2011-01-31 19:37:37 UTC (rev 2635)
@@ -23,6 +23,8 @@
import java.net.MalformedURLException;
import java.net.URL;
+import java.util.Set;
+import java.util.TreeSet;
import org.dllearner.core.ComponentInitException;
import org.dllearner.core.ComponentManager;
@@ -37,9 +39,13 @@
import org.dllearner.kb.KBFile;
import org.dllearner.learningproblems.ClassLearningProblem;
import org.dllearner.learningproblems.Heuristics;
+import org.dllearner.learningproblems.PosNegLPStandard;
import org.dllearner.reasoning.OWLAPIReasoner;
+import org.dllearner.utilities.Helper;
import org.junit.Test;
+import scala.actors.threadpool.Arrays;
+
/**
* Tests for various heuristics employed in learning problems.
*
@@ -159,6 +165,60 @@
}
@Test
+ public void posNegLPLearningTests() throws ComponentInitException {
+ // create artificial ontology
+ KB kb = new KB();
+ String ns = "http://dl-learner.org/junit/";
+ NamedClass[] nc = new NamedClass[5];
+ for(int i=0; i<5; i++) {
+ nc[i] = new NamedClass(ns + "A" + i);
+ }
+ Individual[] ind = new Individual[100];
+ for(int i=0; i<100; i++) {
+ ind[i] = new Individual(ns + "i" + i);
+ }
+
+ // assert individuals to owl:Thing (such that they exist in the knowledge base)
+ for(int i=0; i<100; i++) {
+ kb.addAxiom(new ClassAssertionAxiom(Thing.instance,ind[i]));
+ }
+
+ // A0 has 20 instances (i0 to i19)
+ for(int i=0; i<20; i++) {
+ kb.addAxiom(new ClassAssertionAxiom(nc[0],ind[i]));
+ }
+
+ // A1 has 20 instances (i10 to i29)
+ for(int i=10; i<30; i++) {
+ kb.addAxiom(new ClassAssertionAxiom(nc[1],ind[i]));
+ }
+
+ // A2 has 40 instances (i10 to i49)
+ for(int i=10; i<50; i++) {
+ kb.addAxiom(new ClassAssertionAxiom(nc[2],ind[i]));
+ }
+
+ // A3 has 5 instances (i8 to i12)
+ for(int i=8; i<13; i++) {
+ kb.addAxiom(new ClassAssertionAxiom(nc[3],ind[i]));
+ }
+
+ ComponentManager cm = ComponentManager.getInstance();
+ KnowledgeSource ks = new KBFile(kb);
+ ReasonerComponent reasoner = cm.reasoner(OWLAPIReasoner.class, ks);
+ PosNegLPStandard problem = cm.learningProblem(PosNegLPStandard.class, reasoner);
+ ks.init();
+ reasoner.init();
+
+ Individual[] pos1 = new Individual[] {ind[1], ind[2]};
+ Individual[] neg1 = new Individual[] {ind[3], ind[4]};
+ HeuristicTests.configurePosNegStandardLP(problem, pos1, neg1, "fmeasure", false);
+
+ // TODO: continue
+ }
+
+
+ @Test
public void approximationTests() {
// perform F-Measure example in ontology engineering paper, which was computed on paper
double[] approx1 = Heuristics.getFScoreApproximation(800, 0.8, 1, 10000, 41, 31);
@@ -226,4 +286,19 @@
problem.init();
}
+ @SuppressWarnings("unchecked")
+ private static void configurePosNegStandardLP(PosNegLPStandard problem, Individual[] positiveExamples, Individual[] negativeExamples, String accuracyMethod, boolean useApproximations) throws ComponentInitException {
+ Set<Individual> s1 = new TreeSet<Individual>(Arrays.asList(positiveExamples));
+ Set<Individual> s2 = new TreeSet<Individual>(Arrays.asList(negativeExamples));
+ HeuristicTests.configurePosNegStandardLP(problem, s1, s2, accuracyMethod, useApproximations);
+ }
+
+ // convencience method to set the learning problem to a desired configuration (approximations disabled)
+ private static void configurePosNegStandardLP(PosNegLPStandard problem, Set<Individual> positiveExamples, Set<Individual> negativeExamples, String accuracyMethod, boolean useApproximations) throws ComponentInitException {
+ problem.getConfigurator().setPositiveExamples(Helper.getStringSet(positiveExamples));
+ problem.getConfigurator().setNegativeExamples(Helper.getStringSet(negativeExamples));
+ problem.getConfigurator().setAccuracyMethod(accuracyMethod);
+ problem.getConfigurator().setUseApproximations(useApproximations);
+ problem.init();
+ }
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
