Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[4022] trunk/components-core/src
|
From: <lor...@us...> - 2013-07-29 08:12:38
|
Revision: 4022
http://sourceforge.net/p/dl-learner/code/4022
Author: lorenz_b
Date: 2013-07-29 08:12:30 +0000 (Mon, 29 Jul 2013)
Log Message:
-----------
Cont. ISLE
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LuceneSyntacticIndex.java
trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SimpleSemanticIndex.java
trunk/components-core/src/main/java/org/dllearner/algorithms/qtl/operations/lgg/NoiseSensitiveLGG.java
trunk/components-core/src/main/java/org/dllearner/kb/sparql/QueryEngineHTTP.java
trunk/components-core/src/main/java/org/dllearner/utilities/owl/OWLClassExpressionToSPARQLConverter.java
trunk/components-core/src/test/java/org/dllearner/algorithms/isle/ISLETest.java
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LuceneSyntacticIndex.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LuceneSyntacticIndex.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/LuceneSyntacticIndex.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -67,7 +67,7 @@
} catch (IOException e) {
e.printStackTrace();
}
- return null;
+ return documents;
}
/* (non-Javadoc)
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SimpleSemanticIndex.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SimpleSemanticIndex.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/isle/index/SimpleSemanticIndex.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -3,9 +3,14 @@
*/
package org.dllearner.algorithms.isle.index;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Map.Entry;
import java.util.Set;
+import org.dllearner.algorithms.isle.textretrieval.RDFSLabelEntityTextRetriever;
import org.dllearner.core.owl.Entity;
+import org.semanticweb.owlapi.model.OWLOntology;
/**
* @author Lorenz Buehmann
@@ -13,13 +18,32 @@
*/
public class SimpleSemanticIndex implements SemanticIndex{
+ private SyntacticIndex syntacticIndex;
+ private RDFSLabelEntityTextRetriever labelRetriever;
+ /**
+ *
+ */
+ public SimpleSemanticIndex(OWLOntology ontology, SyntacticIndex syntacticIndex) {
+ this.syntacticIndex = syntacticIndex;
+ labelRetriever = new RDFSLabelEntityTextRetriever(ontology);
+ }
+
+
/* (non-Javadoc)
* @see org.dllearner.algorithms.isle.SemanticIndex#getDocuments(org.dllearner.core.owl.Entity)
*/
@Override
public Set<String> getDocuments(Entity entity) {
- return null;
+ Set<String> documents = new HashSet<String>();
+ 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)
@@ -27,7 +51,7 @@
*/
@Override
public int count(Entity entity) {
- return 0;
+ return getDocuments(entity).size();
}
/* (non-Javadoc)
@@ -35,7 +59,7 @@
*/
@Override
public int getSize() {
- return 0;
+ return syntacticIndex.getSize();
}
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/qtl/operations/lgg/NoiseSensitiveLGG.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/qtl/operations/lgg/NoiseSensitiveLGG.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/qtl/operations/lgg/NoiseSensitiveLGG.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -8,6 +8,7 @@
import java.util.SortedSet;
import java.util.TreeSet;
+import org.apache.log4j.Logger;
import org.dllearner.algorithms.qtl.datastructures.QueryTree;
import org.dllearner.learningproblems.Heuristics;
@@ -16,6 +17,9 @@
public class NoiseSensitiveLGG<N> {
+
+ private static final Logger logger = Logger.getLogger(NoiseSensitiveLGG.class.getName());
+
private LGGGenerator<N> lggGenerator = new LGGGeneratorImpl<N>();
private Queue<EvaluatedQueryTree<N>> todoList;
@@ -29,7 +33,8 @@
Monitor lggMon = MonitorFactory.getTimeMonitor("lgg-mon");
init(trees);
EvaluatedQueryTree<N> currentElement;
- do{System.out.println("TODO list size: " + todoList.size());
+ do{
+ logger.trace("TODO list size: " + todoList.size());
//pick best element from todo list
currentElement = todoList.poll();
for (QueryTree<N> example : currentElement.getUncoveredExamples()) {
@@ -39,33 +44,36 @@
QueryTree<N> lgg = lggGenerator.getLGG(tree, example);
lggMon.stop();
//compute examples which are not covered by LGG
- Collection<QueryTree<N>> uncoveredExamples = new ArrayList<QueryTree<N>>();
- for (QueryTree<N> queryTree : trees) {
- subMon.start();
- boolean subsumed = queryTree.isSubsumedBy(lgg);
- subMon.stop();
- if(!subsumed){
- uncoveredExamples.add(queryTree);
- }
- }
+ Collection<QueryTree<N>> uncoveredExamples = getUncoveredTrees(lgg, trees);
//compute score
double score = Heuristics.getConfidenceInterval95WaldAverage(trees.size(), trees.size() - uncoveredExamples.size());
//add to todo list, if not already contained in todo list or solution list
EvaluatedQueryTree<N> solution = new EvaluatedQueryTree<N>(lgg, uncoveredExamples, score);
todo(solution);
}
- System.out.println("LGG time: " + lggMon.getTotal() + "ms");
- System.out.println("Avg. LGG time: " + lggMon.getAvg() + "ms");
- System.out.println("#LGG computations: " + lggMon.getHits());
- System.out.println("Subsumption test time: " + subMon.getTotal() + "ms");
- System.out.println("Avg. subsumption test time: " + subMon.getAvg() + "ms");
- System.out.println("#Subsumption tests: " + subMon.getHits());
solutions.add(currentElement);
// todoList.remove(currentElement);
} while(!terminationCriteriaSatisfied());
+ logger.trace("LGG time: " + lggMon.getTotal() + "ms");
+ logger.trace("Avg. LGG time: " + lggMon.getAvg() + "ms");
+ logger.trace("#LGG computations: " + lggMon.getHits());
+ logger.trace("Subsumption test time: " + subMon.getTotal() + "ms");
+ logger.trace("Avg. subsumption test time: " + subMon.getAvg() + "ms");
+ logger.trace("#Subsumption tests: " + subMon.getHits());
return new ArrayList<EvaluatedQueryTree<N>>(solutions);
}
+ private Collection<QueryTree<N>> getUncoveredTrees(QueryTree<N> tree, List<QueryTree<N>> allTrees){
+ Collection<QueryTree<N>> uncoveredTrees = new ArrayList<QueryTree<N>>();
+ for (QueryTree<N> queryTree : allTrees) {
+ boolean subsumed = queryTree.isSubsumedBy(tree);
+ if(!subsumed){
+ uncoveredTrees.add(queryTree);
+ }
+ }
+ return uncoveredTrees;
+ }
+
private void init(List<QueryTree<N>> trees){
todoList = new PriorityQueue<EvaluatedQueryTree<N>>();
solutions = new TreeSet<EvaluatedQueryTree<N>>();
@@ -76,9 +84,11 @@
for (QueryTree<N> queryTree : trees) {//System.out.println(queryTree.getStringRepresentation());
boolean distinct = true;
for (QueryTree<N> otherTree : distinctTrees) {
- if(queryTree.isSubsumedBy(otherTree)){
- distinct = false;
- break;
+ if(!queryTree.equals(otherTree)){
+ if(queryTree.isSameTreeAs(otherTree)){
+ distinct = false;
+ break;
+ }
}
}
if(distinct){
@@ -86,9 +96,8 @@
}
}
for (QueryTree<N> queryTree : distinctTrees) {
- Collection<QueryTree<N>> uncoveredExamples = new ArrayList<QueryTree<N>>(distinctTrees);
- uncoveredExamples.remove(queryTree);
- double score = (trees.size() - uncoveredExamples.size()) / (double)trees.size();
+ Collection<QueryTree<N>> uncoveredExamples = getUncoveredTrees(queryTree, trees);
+ double score = Heuristics.getConfidenceInterval95WaldAverage(trees.size(), trees.size() - uncoveredExamples.size());
todoList.add(new EvaluatedQueryTree<N>(queryTree, uncoveredExamples, score));
}
}
Modified: trunk/components-core/src/main/java/org/dllearner/kb/sparql/QueryEngineHTTP.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/kb/sparql/QueryEngineHTTP.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/main/java/org/dllearner/kb/sparql/QueryEngineHTTP.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -268,7 +268,7 @@
private Model execModel(Model model)
{
HttpQuery httpQuery = makeHttpQuery() ;
- httpQuery.setAccept(WebContent.contentTypeNTriplesAlt) ;
+ httpQuery.setAccept(WebContent.contentTypeTurtleAlt1) ;
InputStream in = httpQuery.exec() ;
//Don't assume the endpoint actually gives back the content type we asked for
@@ -284,7 +284,7 @@
//Try to select language appropriately here based on the model content type
Lang lang = WebContent.contentTypeToLang(actualContentType);
if (! RDFLanguages.isTriples(lang)) throw new QueryException("Endpoint returned Content Type: " + actualContentType + " which is not a valid RDF Graph syntax");
- model.read(in, null, Lang.NTRIPLES.getName()) ;
+ model.read(in, null, Lang.TURTLE.getName()) ;
return model ;
}
Modified: trunk/components-core/src/main/java/org/dllearner/utilities/owl/OWLClassExpressionToSPARQLConverter.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/utilities/owl/OWLClassExpressionToSPARQLConverter.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/main/java/org/dllearner/utilities/owl/OWLClassExpressionToSPARQLConverter.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -82,6 +82,8 @@
private Set<? extends OWLEntity> variableEntities = new HashSet<OWLEntity>();
private VariablesMapping mapping;
+ private boolean ignoreGenericTypeStatements = true;
+ private OWLClassExpression expr;
public OWLClassExpressionToSPARQLConverter(VariablesMapping mapping) {
this.mapping = mapping;
@@ -96,6 +98,7 @@
}
public String convert(String rootVariable, OWLClassExpression expr){
+ this.expr = expr;
reset();
variables.push(rootVariable);
expr.accept(this);
@@ -261,7 +264,9 @@
@Override
public void visit(OWLClass ce) {
- sparql += triple(variables.peek(), "a", render(ce));
+ if(ce.equals(expr) || (ignoreGenericTypeStatements && !ce.isOWLThing())){
+ sparql += triple(variables.peek(), "a", render(ce));
+ }
}
@Override
@@ -577,7 +582,9 @@
@Override
public void visit(OWLDatatype node) {
- sparql += "FILTER(DATATYPE(" + variables.peek() + "=<" + node.getIRI().toString() + ">))";
+ if(ignoreGenericTypeStatements && !node.isRDFPlainLiteral() && !node.isTopDatatype()){
+ sparql += "FILTER(DATATYPE(" + variables.peek() + "=<" + node.getIRI().toString() + ">))";
+ }
}
@Override
Modified: trunk/components-core/src/test/java/org/dllearner/algorithms/isle/ISLETest.java
===================================================================
--- trunk/components-core/src/test/java/org/dllearner/algorithms/isle/ISLETest.java 2013-07-17 11:44:41 UTC (rev 4021)
+++ trunk/components-core/src/test/java/org/dllearner/algorithms/isle/ISLETest.java 2013-07-29 08:12:30 UTC (rev 4022)
@@ -6,6 +6,15 @@
import java.io.File;
import java.util.Map;
+import org.dllearner.algorithms.isle.index.OWLOntologyLuceneSyntacticIndexCreator;
+import org.dllearner.algorithms.isle.index.SemanticIndex;
+import org.dllearner.algorithms.isle.index.SimpleSemanticIndex;
+import org.dllearner.algorithms.isle.index.SyntacticIndex;
+import org.dllearner.algorithms.isle.metrics.PMIRelevanceMetric;
+import org.dllearner.algorithms.isle.metrics.RelevanceMetric;
+import org.dllearner.algorithms.isle.metrics.RelevanceUtils;
+import org.dllearner.algorithms.isle.textretrieval.EntityTextRetriever;
+import org.dllearner.algorithms.isle.textretrieval.RDFSLabelEntityTextRetriever;
import org.dllearner.core.AbstractReasonerComponent;
import org.dllearner.core.KnowledgeSource;
import org.dllearner.core.owl.Entity;
@@ -16,9 +25,12 @@
import org.junit.Before;
import org.junit.Test;
import org.semanticweb.owlapi.apibinding.OWLManager;
+import org.semanticweb.owlapi.model.OWLDataFactory;
import org.semanticweb.owlapi.model.OWLOntology;
import org.semanticweb.owlapi.model.OWLOntologyManager;
+import uk.ac.manchester.cs.owl.owlapi.OWLDataFactoryImpl;
+
import com.google.common.base.Joiner;
/**
@@ -29,10 +41,10 @@
private OWLOntologyManager manager;
private OWLOntology ontology;
+ private OWLDataFactory df = new OWLDataFactoryImpl();
private NamedClass cls;
private EntityTextRetriever textRetriever;
- private LuceneSearcher searcher;
- private Relevance relevance;
+ private RelevanceMetric relevance;
private String searchField = "label";
/**
@@ -43,9 +55,9 @@
ontology = manager.loadOntologyFromOntologyDocument(new File("../examples/isle/father_labeled.owl"));
cls = new NamedClass("http://example.com/father#father");
textRetriever = new RDFSLabelEntityTextRetriever(ontology);
- OWLOntologyLuceneIndex index = new OWLOntologyLuceneIndex(ontology, searchField);
- searcher = new LuceneSearcher(index.getDirectory(), searchField);
- relevance = new PMILuceneBasedRelevance(ontology, searcher, textRetriever);
+ SyntacticIndex syntacticIndex = new OWLOntologyLuceneSyntacticIndexCreator(ontology, df.getRDFSLabel(), searchField).buildIndex();
+ SemanticIndex semanticIndex = new SimpleSemanticIndex(ontology, syntacticIndex);
+ relevance = new PMIRelevanceMetric(semanticIndex);
}
/**
@@ -66,7 +78,7 @@
@Test
public void testEntityRelevance() throws Exception {
System.out.println("Relevant entities for entity " + cls + ":");
- Map<Entity, Double> entityRelevance = relevance.getEntityRelevance(cls);
+ Map<Entity, Double> entityRelevance = RelevanceUtils.getRelevantEntities(cls, ontology, relevance);
System.out.println(Joiner.on("\n").join(entityRelevance.entrySet()));
}
@@ -80,7 +92,7 @@
lp.setClassToDescribe(cls);
lp.init();
- Map<Entity, Double> entityRelevance = relevance.getEntityRelevance(cls);
+ Map<Entity, Double> entityRelevance = RelevanceUtils.getRelevantEntities(cls, ontology, relevance);
NLPHeuristic heuristic = new NLPHeuristic(entityRelevance);
ISLE isle = new ISLE(lp, reasoner);
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
