Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3514] trunk/components-core/src/main/java/org/ dllearner
|
From: <lor...@us...> - 2011-12-27 10:37:00
|
Revision: 3514
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3514&view=rev
Author: lorenz_b
Date: 2011-12-27 10:36:53 +0000 (Tue, 27 Dec 2011)
Log Message:
-----------
Continued to get working disjointness algorithm on local model.
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/algorithms/DisjointClassesLearner.java
trunk/components-core/src/main/java/org/dllearner/core/AbstractAxiomLearningAlgorithm.java
trunk/components-core/src/main/java/org/dllearner/kb/LocalModelBasedSparqlEndpointKS.java
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/DisjointClassesLearner.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/DisjointClassesLearner.java 2011-12-21 14:38:40 UTC (rev 3513)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/DisjointClassesLearner.java 2011-12-27 10:36:53 UTC (rev 3514)
@@ -24,6 +24,7 @@
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
+import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
@@ -44,17 +45,21 @@
import org.dllearner.core.owl.DisjointClassesAxiom;
import org.dllearner.core.owl.NamedClass;
import org.dllearner.kb.LocalModelBasedSparqlEndpointKS;
-import org.dllearner.kb.OWLFile;
import org.dllearner.kb.SparqlEndpointKS;
-import org.dllearner.kb.sparql.SPARQLTasks;
import org.dllearner.kb.sparql.SparqlEndpoint;
import org.dllearner.learningproblems.AxiomScore;
import org.dllearner.learningproblems.Heuristics;
+import org.openrdf.model.vocabulary.OWL;
+import org.openrdf.model.vocabulary.RDF;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
+import com.hp.hpl.jena.ontology.OntClass;
+import com.hp.hpl.jena.ontology.OntModel;
import com.hp.hpl.jena.query.QuerySolution;
import com.hp.hpl.jena.query.ResultSet;
+import com.hp.hpl.jena.rdf.model.Model;
+import com.hp.hpl.jena.rdf.model.ModelFactory;
/**
* Learns disjoint classes using SPARQL queries.
@@ -123,64 +128,85 @@
allClasses.remove(classToDescribe);
//get the subclasses
- if(reasoner.isPrepared()){
- subClasses = reasoner.getClassHierarchy().getSubClasses(classToDescribe, false);
+ if(ks.isRemote()){
+ if(reasoner.isPrepared()){
+ subClasses = reasoner.getClassHierarchy().getSubClasses(classToDescribe, false);
+ } else {
+ subClasses = reasoner.getSubClasses(classToDescribe, true);
+ }
} else {
- subClasses = reasoner.getSubClasses(classToDescribe, true);
+ subClasses = new TreeSet<Description>();
+ OntModel ontModel = ((LocalModelBasedSparqlEndpointKS)ks).getModel();
+ OntClass cls = ontModel.getOntClass(classToDescribe.getName());
+ for(OntClass sub : cls.listSubClasses(false).toSet()){
+ if(!sub.isAnon()){
+ subClasses.add(new NamedClass(sub.getURI()));
+ }
+ }
+ for(OntClass sup : cls.listSuperClasses().toSet()){System.out.println(cls.listSuperClasses().toSet());
+ if(!sup.isAnon()){
+ subClasses.add(new NamedClass(sup.getURI()));
+ }
+ }
}
if(ks.supportsSPARQL_1_1()){
runSPARQL1_0_Mode();
} else {
- runSPARQL1_1_Mode();
+ runSPARQL1_0_Mode();
}
- //get classes and how often they occur
-
-
logger.info("...finished in {}ms.", (System.currentTimeMillis()-startTime));
}
private void runSPARQL1_0_Mode(){
+ Model model = ModelFactory.createDefaultModel();
int limit = 1000;
int offset = 0;
- String queryTemplate = "SELECT ?s ?type WHERE {?s a <%s>. ?s a ?type.} LIMIT %d OFFSET %d";
- String query;
+ String baseQuery = "CONSTRUCT {?s a <%s>. ?s a ?type.} WHERE {?s a <%s>. ?s a ?type.} LIMIT %d OFFSET %d";
+ String query = String.format(baseQuery, classToDescribe.getName(), classToDescribe.getName(), limit, offset);System.out.println(query);
+ Model newModel = executeConstructQuery(query);
Map<NamedClass, Integer> result = new HashMap<NamedClass, Integer>();
NamedClass cls;
Integer oldCnt;
- boolean repeat = true;
-
- int total = 0;
-
- String resource = "";
- while(!terminationCriteriaSatisfied() && repeat){
- query = String.format(queryTemplate, classToDescribe, limit, offset);
- ResultSet rs = executeSelectQuery(query);
+ while(newModel.size() != 0){
+ model.add(newModel);
+ //get total number of distinct instances
+ query = "SELECT (COUNT(DISTINCT ?s) AS ?count) WHERE {?s a ?type.}";
+ ResultSet rs = executeSelectQuery(query, model);
+ int total = rs.next().getLiteral("count").getInt();
+
+ // get number of instances of s with <s p o>
+ query = "SELECT ?type (COUNT(?s) AS ?count) WHERE {?s a ?type.}" +
+ " GROUP BY ?type";
+ rs = executeSelectQuery(query, model);
QuerySolution qs;
- repeat = false;
while(rs.hasNext()){
- qs = rs.next();
- String newResource = qs.getResource("?s").getURI();
- if(newResource != resource){
- total++;
- resource = newResource;
+ qs = rs.next();System.out.println(qs);
+ if(qs.getResource("type") != null){
+ cls = new NamedClass(qs.getResource("type").getURI());
+ int newCnt = qs.getLiteral("count").getInt();
+ oldCnt = result.get(cls);
+ if(oldCnt == null){
+ oldCnt = Integer.valueOf(newCnt);
+ } else {
+ oldCnt += newCnt;
+ }
+
+ result.put(cls, oldCnt);
}
- cls = new NamedClass(qs.getResource("type").getURI());
- oldCnt = result.get(cls);
- if(oldCnt == null){
- oldCnt = Integer.valueOf(0);
- }
- int newCnt = oldCnt + 1;
- result.put(cls, newCnt);
- repeat = true;
}
+
if(!result.isEmpty()){
- currentlyBestEvaluatedDescriptions = buildEvaluatedClassDescriptions(result, total);
- offset += 1000;
+ currentlyBestEvaluatedDescriptions = buildEvaluatedClassDescriptions(result, allClasses, total);
}
+
+ offset += limit;
+ query = String.format(baseQuery, classToDescribe.getName(), classToDescribe.getName(), limit, offset);
+ newModel = executeConstructQuery(query);
}
+
}
private void runSPARQL1_1_Mode(){
@@ -201,7 +227,7 @@
QuerySolution qs;
repeat = false;
while(rs.hasNext()){
- qs = rs.next();
+ qs = rs.next();System.out.println(qs);
if(qs.getResource("type") != null){
cls = new NamedClass(qs.getResource("type").getURI());
int newCnt = qs.getLiteral("count").getInt();
@@ -218,7 +244,7 @@
}
if(!result.isEmpty()){
- currentlyBestEvaluatedDescriptions = buildEvaluatedClassDescriptions(result, allClasses);
+ currentlyBestEvaluatedDescriptions = buildEvaluatedClassDescriptions(result, allClasses, result.get(classToDescribe));
offset += 1000;
}
}
@@ -269,11 +295,11 @@
return axioms;
}
- private List<EvaluatedDescription> buildEvaluatedClassDescriptions(Map<NamedClass, Integer> class2Count, Set<NamedClass> allClasses){
+ private List<EvaluatedDescription> buildEvaluatedClassDescriptions(Map<NamedClass, Integer> class2Count, Set<NamedClass> allClasses, int total){
List<EvaluatedDescription> evalDescs = new ArrayList<EvaluatedDescription>();
//Remove temporarily classToDescribe but keep track of their count
- Integer all = class2Count.get(classToDescribe);
+// Integer all = class2Count.get(classToDescribe);
class2Count.remove(classToDescribe);
//get complete disjoint classes
@@ -299,7 +325,12 @@
}
for(NamedClass cls : completeDisjointclasses){
if(useClassPopularity){
- int popularity = reasoner.getIndividualsCount(cls);
+ int popularity = 0;
+ if(ks.isRemote()){
+ popularity = reasoner.getIndividualsCount(cls);
+ } else {
+ popularity = ((LocalModelBasedSparqlEndpointKS)ks).getModel().getOntClass(cls.getName()).listInstances(true).toSet().size();
+ }
//we skip classes with no instances
if(popularity == 0) continue;
double[] confidenceInterval = Heuristics.getConfidenceInterval95Wald(popularity, 0);
@@ -314,66 +345,10 @@
//secondly, create disjoint classexpressions with score 1 - (#occurence/#all)
for(Entry<NamedClass, Integer> entry : sortByValues(class2Count)){
+ //drop classes from OWL and RDF namespace
+ if(entry.getKey().getName().startsWith(OWL.NAMESPACE) || entry.getKey().getName().startsWith(RDF.NAMESPACE))continue;
// evalDesc = new EvaluatedDescription(entry.getKey(),
// new AxiomScore(1 - (entry.getValue() / (double)all)));
- double[] confidenceInterval = Heuristics.getConfidenceInterval95Wald(all, entry.getValue());
- double accuracy = (confidenceInterval[0] + confidenceInterval[1]) / 2;
- evalDesc = new EvaluatedDescription(entry.getKey(),
- new AxiomScore(1 - accuracy));
- evalDescs.add(evalDesc);
- }
-
- class2Count.put(classToDescribe, all);
- return evalDescs;
- }
-
- private List<EvaluatedDescription> buildEvaluatedClassDescriptions(Map<NamedClass, Integer> class2Count, int total){
- List<EvaluatedDescription> evalDescs = new ArrayList<EvaluatedDescription>();
-
- //Remove temporarily classToDescribe but keep track of their count
- class2Count.remove(classToDescribe);
-
- //get complete disjoint classes
- Set<NamedClass> completeDisjointclasses = new TreeSet<NamedClass>(allClasses);
- completeDisjointclasses.removeAll(class2Count.keySet());
-
- //drop all classes which have a super class in this set
- if(suggestMostGeneralClasses && reasoner.isPrepared()){
- keepMostGeneralClasses(completeDisjointclasses);
- }
-
- //we remove the asserted subclasses here
- completeDisjointclasses.removeAll(subClasses);
- for(Description subClass : subClasses){
- class2Count.remove(subClass);
- }
-
-
- EvaluatedDescription evalDesc;
- //firstly, create disjoint classexpressions which not occur and give score of 1
- if(reasoner.isPrepared()){
- SortedSet<Description> mostGeneralClasses = reasoner.getClassHierarchy().getMostGeneralClasses();
- }
- for(NamedClass cls : completeDisjointclasses){
- if(useClassPopularity && (
- (ks instanceof SparqlEndpointKS && ((SparqlEndpointKS) ks).supportsSPARQL_1_1()) || !(ks instanceof SparqlEndpointKS))){
- int popularity = reasoner.getIndividualsCount(cls);
- //we skip classes with no instances
- if(popularity == 0) continue;
- double[] confidenceInterval = Heuristics.getConfidenceInterval95Wald(popularity, 0);
- double accuracy = (confidenceInterval[0] + confidenceInterval[1]) / 2;
- evalDesc = new EvaluatedDescription(cls, new AxiomScore(1- accuracy));
- } else {
- evalDesc = new EvaluatedDescription(cls, new AxiomScore(1));
- }
-
- evalDescs.add(evalDesc);
- }
-
- //secondly, create disjoint classexpressions with score 1 - (#occurence/#all)
- for(Entry<NamedClass, Integer> entry : sortByValues(class2Count)){
-// evalDesc = new EvaluatedDescription(entry.getKey(),
-// new AxiomScore(1 - (entry.getValue() / (double)all)));
double[] confidenceInterval = Heuristics.getConfidenceInterval95Wald(total, entry.getValue());
double accuracy = (confidenceInterval[0] + confidenceInterval[1]) / 2;
evalDesc = new EvaluatedDescription(entry.getKey(),
@@ -385,6 +360,7 @@
return evalDescs;
}
+
private void keepMostGeneralClasses(Set<NamedClass> classes){
ClassHierarchy h = reasoner.getClassHierarchy();
for(NamedClass nc : new HashSet<NamedClass>(classes)){
@@ -393,17 +369,16 @@
}
public static void main(String[] args) throws Exception{
- LocalModelBasedSparqlEndpointKS ks = new LocalModelBasedSparqlEndpointKS(new URL("http://dl-learner.svn.sourceforge.net/viewvc/dl-learner/trunk/examples/swore/swore.rdf?revision=2217"));
- DisjointClassesLearner l = new DisjointClassesLearner(new SparqlEndpointKS(new SparqlEndpoint(new URL("http://dbpedia.aksw.org:8902/sparql"),
- Collections.singletonList("http://dbpedia.org"), Collections.<String>emptyList())));
- l = new DisjointClassesLearner(ks);
+// SparqlEndpointKS ks = new SparqlEndpointKS(new SparqlEndpoint(new URL("http://dbpedia.aksw.org:8902/sparql"), Collections.singletonList("http://dbpedia.org"), Collections.<String>emptyList()));
+ SparqlEndpointKS ks = new LocalModelBasedSparqlEndpointKS(new URL("http://dl-learner.svn.sourceforge.net/viewvc/dl-learner/trunk/examples/swore/swore.rdf?revision=2217"));
+ DisjointClassesLearner l = new DisjointClassesLearner(ks);
l.setClassToDescribe(new NamedClass("http://ns.softwiki.de/req/CustomerRequirement"));
l.init();
- l.getReasoner().prepareSubsumptionHierarchy();
+// l.getReasoner().prepareSubsumptionHierarchy();
// System.out.println(l.getReasoner().getClassHierarchy().getSubClasses(new NamedClass("http://dbpedia.org/ontology/Athlete"), false));System.exit(0);
l.start();
- for(EvaluatedAxiom e : l.getCurrentlyBestEvaluatedAxioms(Integer.MAX_VALUE, 0.75)){
+ for(EvaluatedAxiom e : l.getCurrentlyBestEvaluatedAxioms(Integer.MAX_VALUE, 0.2)){
System.out.println(e);
}
Modified: trunk/components-core/src/main/java/org/dllearner/core/AbstractAxiomLearningAlgorithm.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/core/AbstractAxiomLearningAlgorithm.java 2011-12-21 14:38:40 UTC (rev 3513)
+++ trunk/components-core/src/main/java/org/dllearner/core/AbstractAxiomLearningAlgorithm.java 2011-12-27 10:36:53 UTC (rev 3514)
@@ -45,6 +45,7 @@
import org.dllearner.learningproblems.Heuristics;
import org.dllearner.reasoning.SPARQLReasoner;
import org.dllearner.utilities.owl.AxiomComparator;
+import org.openrdf.model.vocabulary.OWL;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.annotation.Autowired;
@@ -55,6 +56,7 @@
import com.hp.hpl.jena.query.ResultSet;
import com.hp.hpl.jena.rdf.model.Model;
import com.hp.hpl.jena.sparql.engine.http.QueryEngineHTTP;
+import com.hp.hpl.jena.util.iterator.Filter;
/**
* @author Lorenz Bühmann
@@ -193,7 +195,7 @@
return new SPARQLTasks(((SparqlEndpointKS) ks).getEndpoint()).getAllClasses();
} else {
Set<NamedClass> classes = new TreeSet<NamedClass>();
- for(OntClass cls : ((LocalModelBasedSparqlEndpointKS)ks).getModel().listClasses().toList()){
+ for(OntClass cls : ((LocalModelBasedSparqlEndpointKS)ks).getModel().listClasses().filterDrop(new OWLFilter()).toList()){
if(!cls.isAnon()){
classes.add(new NamedClass(cls.getURI()));
}
@@ -235,7 +237,7 @@
}
protected ResultSet executeSelectQuery(String query, Model model) {
- logger.info("Sending query\n{} ...", query);
+ logger.info("Sending query on local model\n{} ...", query);
QueryExecution qexec = QueryExecutionFactory.create(query, model);
ResultSet rs = qexec.execSelect();;
@@ -317,6 +319,17 @@
return new AxiomScore(accuracy, confidence);
}
+ class OWLFilter extends Filter<OntClass>{
+
+ @Override
+ public boolean accept(OntClass cls) {
+ if(!cls.isAnon()){
+ return cls.getURI().startsWith(OWL.NAMESPACE);
+ }
+ return false;
+ }
+
+ }
}
Modified: trunk/components-core/src/main/java/org/dllearner/kb/LocalModelBasedSparqlEndpointKS.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/kb/LocalModelBasedSparqlEndpointKS.java 2011-12-21 14:38:40 UTC (rev 3513)
+++ trunk/components-core/src/main/java/org/dllearner/kb/LocalModelBasedSparqlEndpointKS.java 2011-12-27 10:36:53 UTC (rev 3514)
@@ -35,7 +35,7 @@
// read the RDF/XML file
baseModel.read(in, null);
- model = ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM, baseModel);
+ model = ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM_TRANS_INF, baseModel);
}
@Override
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
