Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[1102] trunk/src/dl-learner/org/dllearner
|
From: <ku...@us...> - 2008-08-19 15:35:27
|
Revision: 1102
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=1102&view=rev
Author: kurzum
Date: 2008-08-19 15:35:22 +0000 (Tue, 19 Aug 2008)
Log Message:
-----------
working improved aquisitor
Modified Paths:
--------------
trunk/src/dl-learner/org/dllearner/kb/aquisitors/LinkedDataTupelAquisitor.java
trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitor.java
trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitorImproved.java
trunk/src/dl-learner/org/dllearner/kb/aquisitors/TupelAquisitor.java
trunk/src/dl-learner/org/dllearner/kb/extraction/ExtractionAlgorithm.java
trunk/src/dl-learner/org/dllearner/kb/manipulator/DBpediaNavigatorOtherRule.java
trunk/src/dl-learner/org/dllearner/kb/manipulator/Manipulator.java
trunk/src/dl-learner/org/dllearner/kb/manipulator/TypeFilterRule.java
trunk/src/dl-learner/org/dllearner/kb/sparql/SparqlKnowledgeSource.java
trunk/src/dl-learner/org/dllearner/test/SparqlExtractionTest.java
Modified: trunk/src/dl-learner/org/dllearner/kb/aquisitors/LinkedDataTupelAquisitor.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/aquisitors/LinkedDataTupelAquisitor.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/aquisitors/LinkedDataTupelAquisitor.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -45,12 +45,17 @@
// standard query get a tupels (p,o) for subject s
@Override
- public SortedSet<RDFNodeTuple> getTupelForResource(String uri) {
-
+ public SortedSet<RDFNodeTuple> retrieveTupel(String uri){
throw new RuntimeException("Not Implemented yet");
-
-
}
+ @Override
+ public SortedSet<RDFNodeTuple> retrieveClassesForInstances(String uri){
+ throw new RuntimeException("Not Implemented yet");
+ }
+ @Override
+ public SortedSet<RDFNodeTuple> retrieveTuplesForClassesOnly(String uri){
+ throw new RuntimeException("Not Implemented yet");
+ }
Modified: trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitor.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitor.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitor.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -34,8 +34,11 @@
*/
public class SparqlTupelAquisitor extends TupelAquisitor {
+
@SuppressWarnings("unused")
private static Logger logger = Logger.getLogger(SparqlTupelAquisitor.class);
+ protected static final String PREDICATE = "predicate";
+ protected static final String OBJECT = "object";
protected SparqlQueryMaker sparqlQueryMaker;
protected SPARQLTasks sparqlTasks;
@@ -48,26 +51,24 @@
this.sparqlTasks = sparqlTasks;
}
-
- // standard query get a tupels (p,o) for subject s
@Override
- public SortedSet<RDFNodeTuple> getTupelForResource(String uri) {
- checkURIforValidity(uri);
+ public SortedSet<RDFNodeTuple> retrieveTupel(String uri){
+ // getQuery
+ String sparqlQueryString = sparqlQueryMaker.makeSubjectQueryUsingFilters(uri);
+ return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, PREDICATE, OBJECT);
- String pred = "predicate";
- String obj = "object";
- String sparqlQueryString = "";
+ }
+ @Override
+ public SortedSet<RDFNodeTuple> retrieveClassesForInstances(String uri){
// getQuery
- if (classMode) {
- sparqlQueryString = sparqlQueryMaker.makeClassQueryUsingFilters(uri);
- }else {
- sparqlQueryString = sparqlQueryMaker.makeSubjectQueryUsingFilters(uri);
- }
+ String sparqlQueryString = sparqlQueryMaker.makeClassQueryUsingFilters(uri);
+ return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, PREDICATE, OBJECT);
-
- return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, pred, obj);
-
}
+ @Override
+ public SortedSet<RDFNodeTuple> retrieveTuplesForClassesOnly(String uri){
+ return retrieveTupel(uri);
+ }
Modified: trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitorImproved.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitorImproved.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/aquisitors/SparqlTupelAquisitorImproved.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -54,61 +54,79 @@
this.recursionDepth = recursionDepth;
}
-
- // standard query get a tupels (p,o) for subject s
-// standard query get a tupels (p,o) for subject s
+
@Override
- public SortedSet<RDFNodeTuple> getTupelForResource(String uri) {
- checkURIforValidity(uri);
- String sparqlQueryString = "";
- String pred = "predicate";
- String obj = "object";
+ public SortedSet<RDFNodeTuple> retrieveTupel(String uri){
- // getQuery
- if (classMode) {
-
-
- sparqlQueryString = sparqlQueryMaker.makeClassQueryUsingFilters(uri);
- return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, pred, obj);
- }
-
SortedSet<RDFNodeTuple> cachedSet = resources.get(uri);
if(cachedSet!=null) {
return cachedSet;
}
//SortedSet<RDFNodeTuple> tmp = new TreeSet<RDFNodeTuple>();
- sparqlQueryString = sparqlQueryMaker.makeSubjectQueryLevel(uri, recursionDepth);
+ String sparqlQueryString = sparqlQueryMaker.makeSubjectQueryLevel(uri, recursionDepth);
+ //System.out.println(sparqlQueryString);
ResultSetRewindable rsw= sparqlTasks.queryAsResultSet(sparqlQueryString);
@SuppressWarnings("unchecked")
List<ResultBinding> l = ResultSetFormatter.toList(rsw);
rsw.reset();
+ int count = 0;
+
+ for (ResultBinding binding : l) {
+ count++;
+ }
+
int resultsetcount = 0;
int i = 0;
for (ResultBinding binding : l) {
- i=0;
- RDFNode nextURI = binding.get(obj+i);
- add(uri, new RDFNodeTuple(binding.get(pred+i), nextURI ));
-
- for (i=1; i < recursionDepth; i++) {
- RDFNode tmpURI = binding.get(obj+i);
- add(nextURI.toString(), new RDFNodeTuple(binding.get(pred+i),tmpURI));
- logger.trace("For: "+nextURI.toString()+ " added :"+resources.get(nextURI.toString()));
- nextURI = tmpURI;
- }
+ i = 0;
+ RDFNode nextOBJ = binding.get(OBJECT+i);
+ RDFNode nextPRED = binding.get(PREDICATE+i);
+ RDFNodeTuple tmptuple = new RDFNodeTuple(nextPRED, nextOBJ );
+ add(uri,tmptuple);
+ boolean cont = !nextOBJ.isLiteral();
+ for (i=0; (i < recursionDepth) && cont; i++) {
+ RDFNode tmpPREDURI = binding.get(PREDICATE+i);
+ RDFNode tmpOBJURI = binding.get(OBJECT+i);
+ if(tmpOBJURI==null) {
+ cont=false;
+ }else if (tmpOBJURI.isLiteral()) {
+ tmptuple = new RDFNodeTuple(tmpPREDURI, tmpOBJURI );
+ add(nextOBJ.toString(), tmptuple);
+ logger.trace(tmptuple);
+ logger.trace("For: "+nextOBJ.toString()+ " added :"+resources.get(nextOBJ.toString()));
+ cont=false;
+ }else {
+ tmptuple = new RDFNodeTuple(tmpPREDURI, tmpOBJURI );
+ add(nextOBJ.toString(), tmptuple);
+ logger.trace(tmptuple);
+ logger.trace("For: "+nextOBJ.toString()+ " added :"+resources.get(nextOBJ.toString()));
+ nextOBJ = tmpOBJURI;
+ cont = true;
+ }
+ }//end for
+
resultsetcount++;
}
+ //System.out.println("original count "+count);
+ logger.warn("SparqlTupelAquisitor retrieved : "+resultsetcount);
if(resultsetcount>999) {
logger.warn("SparqlTupelAquisitor retrieved more than 1000 results, there might some be missing");
}
- return resources.get(uri);
-
- //return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, pred, obj);
+ return ((cachedSet=resources.get(uri))==null)?new TreeSet<RDFNodeTuple>():cachedSet;
}
+ @Override
+ public SortedSet<RDFNodeTuple> retrieveTuplesForClassesOnly(String uri){
+ //getQuery
+ String sparqlQueryString = sparqlQueryMaker.makeSubjectQueryUsingFilters(uri);
+ return sparqlTasks.queryAsRDFNodeTuple(sparqlQueryString, PREDICATE, OBJECT);
+ }
+
+
private void add(String uri, RDFNodeTuple tuple){
SortedSet<RDFNodeTuple> set = resources.get(uri);
if(set==null){
Modified: trunk/src/dl-learner/org/dllearner/kb/aquisitors/TupelAquisitor.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/aquisitors/TupelAquisitor.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/aquisitors/TupelAquisitor.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -38,20 +38,41 @@
private static Logger logger = Logger.getLogger(TupelAquisitor.class);
+ protected final int NORMAL = 0;
+ protected final int CLASSES_FOR_INSTANCES = 1;
+ protected final int CLASS_INFORMATION = 2;
- protected boolean classMode = false;
+ protected int mode = 0;
private boolean uriDebugCheck = true;
- public abstract SortedSet<RDFNodeTuple> getTupelForResource(String uri);
+ public final SortedSet<RDFNodeTuple> getTupelForResource(String uri){
+ checkURIforValidity(uri);
+ if (mode == NORMAL) {
+ return retrieveTupel(uri);
+ } else if(mode == CLASSES_FOR_INSTANCES){
+ return retrieveClassesForInstances(uri);
+ }else if(mode == CLASS_INFORMATION){
+ return retrieveTuplesForClassesOnly(uri);
+ }else{
+ throw new RuntimeException("undefined mode in aquisitor");
+ }
+ }
+ public abstract SortedSet<RDFNodeTuple> retrieveTupel(String uri);
+ public abstract SortedSet<RDFNodeTuple> retrieveClassesForInstances(String uri);
+ public abstract SortedSet<RDFNodeTuple> retrieveTuplesForClassesOnly(String uri);
- public void setClassMode(boolean classMode) {
- this.classMode = classMode;
- }
+ /*private void setMode(int mode) {
+ this.mode = mode;
+ }*/
- public boolean isClassMode() {
- return classMode;
+ public int getMode() {
+ return mode;
}
+ public void setNextTaskToNormal(){mode = NORMAL;}
+ public void setNextTaskToClassesForInstances(){mode = CLASSES_FOR_INSTANCES;}
+ public void setNextTaskToClassInformation(){mode = CLASS_INFORMATION;}
+
protected boolean checkURIforValidity(String uri){
if(uriDebugCheck) return true;
try{
Modified: trunk/src/dl-learner/org/dllearner/kb/extraction/ExtractionAlgorithm.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/extraction/ExtractionAlgorithm.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/extraction/ExtractionAlgorithm.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -70,6 +70,7 @@
public Node expandNode(String uri, TupelAquisitor tupelAquisitor) {
SimpleClock sc = new SimpleClock();
+ tupelAquisitor.setNextTaskToNormal();
Node seedNode = getFirstNode(uri);
List<Node> newNodes = new ArrayList<Node>();
@@ -131,7 +132,7 @@
//TODO LinkedData incompatibility
- tupelAquisitor.setClassMode(true);
+ tupelAquisitor.setNextTaskToClassesForInstances();
if (configuration.isCloseAfterRecursion()) {
while (!instances.isEmpty()) {
logger.trace("Getting classes for remaining instances: "
@@ -144,12 +145,11 @@
}//endif
}//endwhile
}//endif
- tupelAquisitor.setClassMode(false);
-
+ tupelAquisitor.setNextTaskToClassInformation();
int i = 0;
- while (!classes.isEmpty()) {
+ while (!classes.isEmpty() && false) {
logger.trace("Remaining classes: " + classes.size());
Node next = classes.remove(0);
if (!alreadyQueriedSuperClasses.contains(next.getURI().toString())) {
Modified: trunk/src/dl-learner/org/dllearner/kb/manipulator/DBpediaNavigatorOtherRule.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/manipulator/DBpediaNavigatorOtherRule.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/manipulator/DBpediaNavigatorOtherRule.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -51,7 +51,6 @@
typeTuple = tuple;
}
- //TODO this doesn't work, because it is unclear what toString() method returns
if (tuple.a.toString().equals("http://www.w3.org/2003/01/geo/wgs84_pos#lat") && tuple.b.isLiteral()){
lat = ((Literal) tuple.b).getFloat();
Modified: trunk/src/dl-learner/org/dllearner/kb/manipulator/Manipulator.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/manipulator/Manipulator.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/manipulator/Manipulator.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -23,6 +23,7 @@
import java.util.List;
import java.util.SortedSet;
+import org.apache.log4j.Logger;
import org.dllearner.kb.extraction.ClassNode;
import org.dllearner.kb.extraction.InstanceNode;
import org.dllearner.kb.extraction.Node;
@@ -39,6 +40,7 @@
*/
public class Manipulator {
+ private static Logger logger = Logger.getLogger(Manipulator.class);
private List<Rule> rules = new ArrayList<Rule>();
private Manipulator() {
@@ -56,9 +58,11 @@
*/
public SortedSet<RDFNodeTuple> manipulate( Node node, SortedSet<RDFNodeTuple> tuples) {
JamonMonitorLogger.getTimeMonitor(Manipulator.class, "Time for Rules").start();
+ logger.warn("before: "+tuples.size());
for (Rule rule : rules) {
tuples = rule.applyRule(node, tuples);
}
+ logger.warn("after: "+tuples.size());
JamonMonitorLogger.getTimeMonitor(Manipulator.class, "Time for Rules").stop();
return tuples;
}
@@ -111,9 +115,13 @@
private void addDefaultRules(Months month){
- addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_CLASS,ClassNode.class.getCanonicalName() )) ;
- addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_THING,InstanceNode.class.getCanonicalName() )) ;
- addRule(new TypeFilterRule(month, "", OWLVocabulary.OWL_CLASS, ClassNode.class.getCanonicalName()) ) ;
+ // addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_CLASS,ClassNode.class.getCanonicalName() )) ;
+ // addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_THING,InstanceNode.class.getCanonicalName() )) ;
+ // addRule(new TypeFilterRule(month, "", OWLVocabulary.OWL_CLASS, ClassNode.class.getCanonicalName()) ) ;
+ addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_CLASS,ClassNode.class )) ;
+ addRule(new TypeFilterRule(month, OWLVocabulary.RDF_TYPE, OWLVocabulary.OWL_THING,InstanceNode.class )) ;
+ addRule(new TypeFilterRule(month, "", OWLVocabulary.OWL_CLASS, ClassNode.class) ) ;
+
}
public synchronized void addRule(Rule newRule){
Modified: trunk/src/dl-learner/org/dllearner/kb/manipulator/TypeFilterRule.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/manipulator/TypeFilterRule.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/manipulator/TypeFilterRule.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -22,21 +22,24 @@
import java.util.SortedSet;
import java.util.TreeSet;
+import org.apache.log4j.Logger;
import org.dllearner.kb.extraction.Node;
import org.dllearner.utilities.datastructures.RDFNodeTuple;
public class TypeFilterRule extends Rule{
+ public static Logger logger = Logger.getLogger(TypeFilterRule.class);
+
String predicateFilter;
String objectFilter;
- String canonicalClassName;
+ String classCanonicalName;
- public TypeFilterRule(Months month, String predicateFilter, String objectFilter, String canonicalClassName) {
+ public TypeFilterRule(Months month, String predicateFilter, String objectFilter, Class<? extends Node> clazz) {
super(month);
this.predicateFilter = predicateFilter;
this.objectFilter = objectFilter;
- this.canonicalClassName = canonicalClassName;
+ this.classCanonicalName = clazz.getCanonicalName();
}
@@ -45,12 +48,20 @@
public SortedSet<RDFNodeTuple> applyRule(Node subject, SortedSet<RDFNodeTuple> tuples){
SortedSet<RDFNodeTuple> keep = new TreeSet<RDFNodeTuple>();
for (RDFNodeTuple tuple : tuples) {
+ String a = tuple.a.toString();
+ String b = tuple.b.toString();
+ //System.out.println(a+b);
boolean remove = (tuple.aPartContains(predicateFilter) &&
- tuple.bPartContains(objectFilter) &&
- subject.getClass().getCanonicalName().equals(canonicalClassName));
+ tuple.bPartContains(objectFilter) &&
+ // QUALITY this might be dead wrong
+ (classCanonicalName.equalsIgnoreCase(subject.getClass().getCanonicalName()))
+ );
if(!remove){
keep.add(tuple);
+ }else{
+ logger.warn("Removed: "+subject+"::"+tuple);
}
+
}
return keep;
}
Modified: trunk/src/dl-learner/org/dllearner/kb/sparql/SparqlKnowledgeSource.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/sparql/SparqlKnowledgeSource.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/kb/sparql/SparqlKnowledgeSource.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -47,6 +47,7 @@
import org.dllearner.core.config.StringTupleListConfigOption;
import org.dllearner.core.owl.KB;
import org.dllearner.kb.aquisitors.SparqlTupelAquisitor;
+import org.dllearner.kb.aquisitors.SparqlTupelAquisitorImproved;
import org.dllearner.kb.aquisitors.TupelAquisitor;
import org.dllearner.kb.extraction.Configuration;
import org.dllearner.kb.extraction.Manager;
@@ -410,7 +411,8 @@
public TupelAquisitor getTupelAquisitor()
{
- return new SparqlTupelAquisitor(getSparqlQueryMaker(), getSPARQLTasks());
+ //return new SparqlTupelAquisitor(getSparqlQueryMaker(), getSPARQLTasks());
+ return new SparqlTupelAquisitorImproved(getSparqlQueryMaker(), getSPARQLTasks(),recursionDepth);
}
/* (non-Javadoc)
Modified: trunk/src/dl-learner/org/dllearner/test/SparqlExtractionTest.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/test/SparqlExtractionTest.java 2008-08-19 15:23:53 UTC (rev 1101)
+++ trunk/src/dl-learner/org/dllearner/test/SparqlExtractionTest.java 2008-08-19 15:35:22 UTC (rev 1102)
@@ -26,7 +26,7 @@
import org.apache.log4j.Level;
import org.apache.log4j.Logger;
import org.apache.log4j.SimpleLayout;
-import org.dllearner.kb.aquisitors.SparqlTupelAquisitor;
+import org.dllearner.kb.aquisitors.SparqlTupelAquisitorImproved;
import org.dllearner.kb.extraction.Configuration;
import org.dllearner.kb.extraction.Manager;
import org.dllearner.kb.manipulator.Manipulator;
@@ -34,6 +34,7 @@
import org.dllearner.kb.sparql.SparqlQueryMaker;
import org.dllearner.scripts.NT2RDF;
import org.dllearner.utilities.JamonMonitorLogger;
+import org.dllearner.utilities.statistics.SimpleClock;
/**
* Test class, uses the whole thing
@@ -55,17 +56,19 @@
ConsoleAppender consoleAppender = new ConsoleAppender(layout);
logger.removeAllAppenders();
logger.addAppender(consoleAppender);
- logger.setLevel(Level.TRACE);
+ logger.setLevel(Level.INFO);
// String test2 = "http://www.extraction.org/config#dbpediatest";
// String test = "http://www.extraction.org/config#localjoseki";
try {
// URI u = new URI(test);
+ int recursionDepth=3;
Manager m = new Manager();
Configuration conf = new Configuration (
- new SparqlTupelAquisitor(SparqlQueryMaker.getTestFilter(), SPARQLTasks.getPredefinedSPARQLTasksWithCache("DBPEDIA")),
+ new SparqlTupelAquisitorImproved(SparqlQueryMaker.getAllowYAGOFilter(),
+ SPARQLTasks.getPredefinedSPARQLTasksWithCache("DBPEDIA"),recursionDepth),
Manipulator.getDefaultManipulator(),
- 1,
+ recursionDepth,
true,
true,
200
@@ -79,6 +82,7 @@
fw.write(m.extract(u2));
fw.flush();
fw.close();
+
NT2RDF.convertNT2RDF(filename);
JamonMonitorLogger.printAllSortedByLabel();
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
