Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner: [830] trunk/src/dl-learner/org/dllearner
|
From: <ku...@us...> - 2008-05-06 18:16:31
|
Revision: 830
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=830&view=rev
Author: kurzum
Date: 2008-05-06 11:16:26 -0700 (Tue, 06 May 2008)
Log Message:
-----------
script for learning roles automatically, conf files are created in a folder examples/stest, which must exist
Modified Paths:
--------------
trunk/src/dl-learner/org/dllearner/algorithms/refinement/ROLearner.java
trunk/src/dl-learner/org/dllearner/kb/sparql/configuration/SparqlEndpoint.java
trunk/src/dl-learner/org/dllearner/scripts/SPARQLMassLearning.java
trunk/src/dl-learner/org/dllearner/test/SparqlEndpointTest.java
trunk/src/dl-learner/org/dllearner/utilities/ConfWriter.java
Added Paths:
-----------
trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderRolesSPARQL.java
trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java
Removed Paths:
-------------
trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java
Modified: trunk/src/dl-learner/org/dllearner/algorithms/refinement/ROLearner.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/algorithms/refinement/ROLearner.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/algorithms/refinement/ROLearner.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -567,9 +567,11 @@
if(show>=5){break;}
show++;
}
+
}
logger.info(" horizontal expansion: " + minimumHorizontalExpansion + " to " + maximumHorizontalExpansion);
logger.info(" size of candidate set: " + candidates.size());
+
//logger.trace("test");
//logger.trace(solutions.size());
printBestSolutions(0);
Modified: trunk/src/dl-learner/org/dllearner/kb/sparql/configuration/SparqlEndpoint.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/kb/sparql/configuration/SparqlEndpoint.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/kb/sparql/configuration/SparqlEndpoint.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -291,7 +291,9 @@
} catch (Exception e) {
e.printStackTrace();
}
- return new SparqlEndpoint(u);
+ LinkedList<String> defaultGraphURIs=new LinkedList<String>();
+ defaultGraphURIs.add("http://www.rdfabout.com/rdf/schema/census/");
+ return new SparqlEndpoint(u, defaultGraphURIs, new LinkedList<String>());
}
Modified: trunk/src/dl-learner/org/dllearner/scripts/SPARQLMassLearning.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/scripts/SPARQLMassLearning.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/scripts/SPARQLMassLearning.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -11,6 +11,7 @@
import org.dllearner.kb.sparql.Cache;
import org.dllearner.kb.sparql.SparqlKnowledgeSource;
import org.dllearner.kb.sparql.configuration.SparqlEndpoint;
+import org.dllearner.utilities.AutomaticExampleFinderRolesSPARQL;
import org.dllearner.utilities.AutomaticExampleFinderSPARQL;
import org.dllearner.utilities.ConfWriter;
import org.dllearner.utilities.SimpleClock;
@@ -20,82 +21,168 @@
static Cache c;
static SparqlEndpoint se;
private static Logger logger = Logger.getRootLogger();
+
+ static String standardSettings="";
+ static String algorithm="refexamples";
+ static String standardSettingsRefexamples =
+ "refexamples.minExecutionTimeInSeconds = 30;\n" +
+ "refexamples.maxExecutionTimeInSeconds = 30;\n" +
+ "//refexamples.guaranteeXgoodDescriptions = 10;\n" +
+ "refexamples.logLevel=\"TRACE\";\n" +
+ "refexamples.noisePercentage = 0.10;\n" +
+ "refexamples.writeSearchTree = false;\n" +
+ "refexamples.searchTreeFile = \"searchTree.txt\";\n" +
+ "refexamples.replaceSearchTree = true;\n\n" ;
+
+ static String standardSettingsRefinement =
+ "refinement.minExecutionTimeInSeconds = 30;\n" +
+ "refinement.maxExecutionTimeInSeconds = 30;\n" +
+ "//refinement.guaranteeXgoodDescriptions = 10;\n" +
+ "refinement.logLevel=\"TRACE\";\n" +
+ "refinement.writeSearchTree = false;\n" +
+ "refinement.searchTreeFile = \"searchTree.txt\";\n" +
+ "refinement.replaceSearchTree = true;\n\n" ;
+
+
+
+ static String standardDBpedia="" +
+ "sparql.recursionDepth = 1;\n" +
+ "sparql.predefinedFilter = \"YAGO\";\n" +
+ "sparql.predefinedEndpoint = \"DBPEDIA\";\n";
+ //"sparql.logLevel = \"INFO\";\n";
+
+
+ //vars
+ static boolean useRelated = false;
+ static boolean useSuperClasses = false;
+ static boolean useParallelClasses = true;
+ static int poslimit = 10;
+ static int neglimit = 20;
+
/**
* @param args
*/
public static void main(String[] args) {
init();
+ //logger.setLevel(Level.TRACE);
+ Logger.getLogger(SparqlKnowledgeSource.class).setLevel(Level.INFO);
+ //System.out.println(Logger.getLogger(SparqlQuery.class).getLevel());
+ SimpleClock sc=new SimpleClock();
- //vars
- boolean useRelated = false;
- boolean useSuperClasses = false;
- boolean useParallelClasses = true;
- int poslimit = 10;
- int neglimit = 20;
+ standardSettings=standardSettingsRefexamples+standardDBpedia;
+ standardSettings=standardSettingsRefinement+standardDBpedia;
+ //DBpedia();
+ //algorithm="refinement";
+ roles();
-
- try {
- //logger.setLevel(Level.TRACE);
- Logger.getLogger(SparqlKnowledgeSource.class).setLevel(Level.INFO);
- //System.out.println(Logger.getLogger(SparqlQuery.class).getLevel());
-
- /*System.out.println(Level.DEBUG.getClass());
+ /*System.out.println(Level.DEBUG.getClass());
System.out.println(Level.toLevel("INFO"));
System.out.println(Level.INFO);*/
//System.exit(0);
- SimpleClock sc=new SimpleClock();
- //concepts.add("(EXISTS \"monarch\".TOP AND EXISTS \"predecessor\".(\"Knight\" OR \"Secretary\"))");
-
- SortedSet<String> concepts = new TreeSet<String>();
- concepts.add("(\"http://dbpedia.org/class/yago/HeadOfState110164747\" AND (\"http://dbpedia.org/class/yago/Negotiator110351874\" AND \"http://dbpedia.org/class/yago/Representative110522035\"))");
- //concepts.add("\"http://dbpedia.org/class/yago/Person100007846\"");
- //concepts.add("\"http://dbpedia.org/class/yago/FieldMarshal110086821\"");
- //concepts.add("http://dbpedia.org/resource/Category:Prime_Ministers_of_the_United_Kingdom");
- //concepts.add("http://dbpedia.org/resource/Category:Grammy_Award_winners");
- //concepts.add("EXISTS \"http://dbpedia.org/property/grammyawards\".TOP");
- SortedSet<String> posExamples = new TreeSet<String>();
- SortedSet<String> negExamples = new TreeSet<String>();
- String url = "http://dbpedia.openlinksw.com:8890/sparql";
- //HashMap<String, ResultSet> result = new HashMap<String, ResultSet>();
- //HashMap<String, String> result2 = new HashMap<String, String>();
- //System.out.println(concepts.first());
- //logger.setLevel(Level.TRACE);
- AutomaticExampleFinderSPARQL ae= new AutomaticExampleFinderSPARQL( se);
- //System.out.println(new JenaResultSetConvenience(ae.queryConcept(concepts.first(), 0)).getStringListForVariable("?subject") );;
- //System.out.println(new JenaResultSetConvenience(ae.queryConcept(concepts.first(), 0)).getStringListForVariable("?subject").size() );;
- //ae.getSubClasses(concepts.first());
- //System.exit(0);
-
- ae.init(concepts.first(), useRelated, useSuperClasses,useParallelClasses, poslimit, neglimit);
+
- posExamples = ae.getPosExamples();
- negExamples = ae.getNegExamples();
-
- System.out.println(posExamples);
- System.out.println(negExamples);
- //System.exit(0);
- String tmp = concepts.first().replace("http://dbpedia.org/resource/Category:", "").replace("\"","");
- tmp = tmp.replace("http://dbpedia.org/class/yago/", "");
- tmp = tmp.replace("http://dbpedia.org/property/", "");
- String confname = URLEncoder.encode(tmp, "UTF-8")+".conf";
- //
- ConfWriter cf=new ConfWriter();
- cf.addToStats("relearned concept: "+concepts.first());
-
- //System.exit(0);
- //"relearned concept: ";
- cf.writeSPARQL(confname, posExamples, negExamples, url, new TreeSet<String>());
- //new LearnSparql().learn(posExamples, negExamples, "http://dbpedia.openlinksw.com:8890/sparql", new TreeSet<String>());
-
sc.printAndSet("Finished");
- } catch (Exception e) {
- e.printStackTrace();
- }
-
}
+
+ static void DBpedia(){
+ se = SparqlEndpoint.EndpointDBpedia();
+ //concepts.add("(EXISTS \"monarch\".TOP AND EXISTS \"predecessor\".(\"Knight\" OR \"Secretary\"))");
+
+ SortedSet<String> concepts = new TreeSet<String>();
+ concepts.add("(\"http://dbpedia.org/class/yago/HeadOfState110164747\" AND (\"http://dbpedia.org/class/yago/Negotiator110351874\" AND \"http://dbpedia.org/class/yago/Representative110522035\"))");
+ //concepts.add("\"http://dbpedia.org/class/yago/Person100007846\"");
+ //concepts.add("\"http://dbpedia.org/class/yago/FieldMarshal110086821\"");
+ //concepts.add("http://dbpedia.org/resource/Category:Prime_Ministers_of_the_United_Kingdom");
+ //concepts.add("http://dbpedia.org/resource/Category:Grammy_Award_winners");
+ //concepts.add("EXISTS \"http://dbpedia.org/property/grammyawards\".TOP");
+
+ SortedSet<String> posExamples = new TreeSet<String>();
+ SortedSet<String> negExamples = new TreeSet<String>();
+ String url = "http://dbpedia.openlinksw.com:8890/sparql";
+ //HashMap<String, ResultSet> result = new HashMap<String, ResultSet>();
+ //HashMap<String, String> result2 = new HashMap<String, String>();
+ //System.out.println(concepts.first());
+ //logger.setLevel(Level.TRACE);
+ AutomaticExampleFinderSPARQL ae= new AutomaticExampleFinderSPARQL( se);
+ //System.out.println(new JenaResultSetConvenience(ae.queryConcept(concepts.first(), 0)).getStringListForVariable("?subject") );;
+ //System.out.println(new JenaResultSetConvenience(ae.queryConcept(concepts.first(), 0)).getStringListForVariable("?subject").size() );;
+ //ae.getSubClasses(concepts.first());
+ //System.exit(0);
+
+ ae.initDBpedia(concepts.first(), useRelated, useSuperClasses,useParallelClasses, poslimit, neglimit);
+
+ posExamples = ae.getPosExamples();
+ negExamples = ae.getNegExamples();
+
+ System.out.println(posExamples);
+ System.out.println(negExamples);
+ //System.exit(0);
+ String tmp = concepts.first().replace("http://dbpedia.org/resource/Category:", "").replace("\"","");
+ tmp = tmp.replace("http://dbpedia.org/class/yago/", "");
+ tmp = tmp.replace("http://dbpedia.org/property/", "");
+ String confname = "";
+ try{
+ confname = URLEncoder.encode(tmp, "UTF-8")+".conf";
+ }catch (Exception e) {e.printStackTrace();}
+ //
+ ConfWriter cf=new ConfWriter();
+ cf.addToStats("relearned concept: "+concepts.first());
+
+ //System.exit(0);
+ //"relearned concept: ";
+ cf.writeSPARQL(confname, posExamples, negExamples, url, new TreeSet<String>(),standardSettings,algorithm);
+ //new LearnSparql().learn(posExamples, negExamples, "http://dbpedia.openlinksw.com:8890/sparql", new TreeSet<String>());
+
+
+ }
+
+
+
+ static void roles(){
+
+ se = SparqlEndpoint.EndpointDBpedia();
+ //se = SparqlEndpoint.EndpointUSCensus();
+ SortedSet<String> roles = new TreeSet<String>();
+ roles.add("http://dbpedia.org/property/birthPlace");
+ //roles.add("http://www.rdfabout.com/rdf/schema/census/landArea");
+ standardSettings+=algorithm+".ignoredRoles = {\""+roles.first()+"\"};\n";
+
+ SortedSet<String> posExamples = new TreeSet<String>();
+ SortedSet<String> negExamples = new TreeSet<String>();
+ String url = "http://dbpedia.openlinksw.com:8890/sparql";
+ //HashMap<String, ResultSet> result = new HashMap<String, ResultSet>();
+ //HashMap<String, String> result2 = new HashMap<String, String>();
+ //System.out.println(concepts.first());
+ //logger.setLevel(Level.TRACE);
+ AutomaticExampleFinderRolesSPARQL ae= new AutomaticExampleFinderRolesSPARQL( se);
+
+ ae.initDomainRange(roles.first(), poslimit, neglimit);
+
+ posExamples = ae.getPosExamples();
+ negExamples = ae.getNegExamples();
+
+ System.out.println(posExamples);
+ System.out.println(negExamples);
+ //System.exit(0);
+ String tmp = roles.first().replace("http://dbpedia.org/property/", "").replace("\"","");
+ String confname = "";
+ try{
+ confname = URLEncoder.encode(tmp, "UTF-8")+".conf";
+ }catch (Exception e) {e.printStackTrace();}
+ //
+ ConfWriter cf=new ConfWriter();
+ cf.addToStats("relearned role: "+roles.first());
+
+ //System.exit(0);
+ //"relearned concept: ";
+ cf.writeSPARQL(confname, posExamples, negExamples, url, new TreeSet<String>(),standardSettings,algorithm);
+ //new LearnSparql().learn(posExamples, negExamples, "http://dbpedia.openlinksw.com:8890/sparql", new TreeSet<String>());
+
+
+ }
/***************************************************************************
* *********************OLDCODE String
@@ -131,7 +218,7 @@
public static void init() {
- se = SparqlEndpoint.EndpointDBpedia();
+
// create logger (a simple logger which outputs
// its messages to the console)
SimpleLayout layout = new SimpleLayout();
Modified: trunk/src/dl-learner/org/dllearner/test/SparqlEndpointTest.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/test/SparqlEndpointTest.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/test/SparqlEndpointTest.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -5,6 +5,7 @@
import java.util.Set;
import org.apache.log4j.ConsoleAppender;
+import org.apache.log4j.FileAppender;
import org.apache.log4j.Level;
import org.apache.log4j.Logger;
import org.apache.log4j.SimpleLayout;
@@ -24,7 +25,12 @@
public static void main(String[] args) {
SimpleLayout layout = new SimpleLayout();
ConsoleAppender consoleAppender = new ConsoleAppender(layout);
+ FileAppender fileAppender =null; ;
+ try{
+ fileAppender = new FileAppender(layout,"endpoints.txt",false);
+ }catch (Exception e) {e.printStackTrace();}
logger.removeAllAppenders();
+ logger.addAppender(fileAppender);
logger.addAppender(consoleAppender);
logger.setLevel(Level.DEBUG);
Logger.getLogger(SparqlKnowledgeSource.class).setLevel(Level.INFO);
Added: trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderRolesSPARQL.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderRolesSPARQL.java (rev 0)
+++ trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderRolesSPARQL.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -0,0 +1,123 @@
+package org.dllearner.utilities;
+
+import java.util.SortedSet;
+import java.util.TreeSet;
+
+import org.apache.log4j.Logger;
+import org.dllearner.core.ComponentManager;
+import org.dllearner.kb.sparql.Cache;
+import org.dllearner.kb.sparql.SparqlQuery;
+import org.dllearner.kb.sparql.configuration.SparqlEndpoint;
+
+import com.hp.hpl.jena.query.ResultSet;
+
+public class AutomaticExampleFinderRolesSPARQL {
+
+ private static Logger logger = Logger
+ .getLogger(ComponentManager.class);
+
+ private Cache c;
+ private SparqlEndpoint se;
+ private SortedSet<String> posExamples;
+ private SortedSet<String> negExamples;
+ private int roleLimit=1000;
+
+
+ public AutomaticExampleFinderRolesSPARQL(SparqlEndpoint se){
+ this.c=new Cache();
+ this.se=se;
+ posExamples = new TreeSet<String>();
+ negExamples = new TreeSet<String>();
+ }
+
+ public void initDomainRange(String role, int poslimit, int neglimit) {
+ makePositiveExamplesAsDomain( role);
+ SortedSet<String> keepForClean = new TreeSet<String>();
+ keepForClean.addAll(this.posExamples);
+ this.posExamples = SetManipulation.fuzzyShrink(this.posExamples, poslimit);
+ logger.trace("shrinking: pos Example size: "+posExamples.size());
+
+ makeNegativeExamplesAsRange( role);
+
+
+ //clean
+ negExamples.removeAll(keepForClean);
+ logger.trace("neg Example size after cleaning: "+negExamples.size());
+ this.negExamples = SetManipulation.fuzzyShrink(negExamples, neglimit);
+ logger.debug("pos Example size after shrinking: "+posExamples.size());
+ logger.debug("neg Example size after shrinking: "+negExamples.size());
+ logger.debug("Finished examples for role: "+role);
+ }
+
+
+
+
+ private void makePositiveExamplesAsDomain(String role){
+ logger.debug("making Positive Examples from Role as Domain: "+role);
+ this.posExamples.addAll(getDomain( role, roleLimit));
+ logger.debug(" pos Example size: "+posExamples.size());
+ }
+
+ private void makeNegativeExamplesAsRange(String role){
+ logger.debug("making Negative Examples from Role as Range: "+role);
+ this.negExamples.addAll(getRange( role, roleLimit));
+ logger.debug(" neg Example size: "+negExamples.size());
+ }
+
+ private SortedSet<String> getDomain(String role,int limit){
+
+ String query = "" +
+ "SELECT DISTINCT ?domain " +
+ "WHERE { \n" +
+ "?domain <" + role + "> " + " ?o. \n" +
+ "?domain a []\n." +
+ "FILTER (!isLiteral(?domain))." +
+ "}\n" +
+ "LIMIT "+limit;
+
+
+ String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
+ ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
+ JenaResultSetConvenience rsc = new JenaResultSetConvenience(rs);
+ return rsc.getStringListForVariable("domain");
+
+ }
+
+ private SortedSet<String> getRange(String role,int limit){
+
+ String query = "" +
+ "SELECT DISTINCT ?range " +
+ "WHERE { \n" +
+ "?s <" + role + "> " + " ?range. \n" +
+ "?range a [].\n" +
+ "FILTER (!isLiteral(?range))." +
+ "}\n" +
+ "LIMIT "+limit;
+
+
+ String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
+ ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
+ JenaResultSetConvenience rsc = new JenaResultSetConvenience(rs);
+ return rsc.getStringListForVariable("range");
+
+ }
+
+
+
+
+
+
+ public SortedSet<String> getPosExamples() {
+ return posExamples;
+ }
+
+ public SortedSet<String> getNegExamples() {
+ return negExamples;
+ }
+
+
+
+
+
+
+}
Deleted: trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -1,294 +0,0 @@
-package org.dllearner.utilities;
-
-import java.util.SortedSet;
-import java.util.TreeSet;
-
-import org.apache.log4j.Logger;
-import org.dllearner.core.ComponentManager;
-import org.dllearner.kb.sparql.Cache;
-import org.dllearner.kb.sparql.SparqlQuery;
-import org.dllearner.kb.sparql.SparqlQueryDescriptionConvertVisitor;
-import org.dllearner.kb.sparql.configuration.SparqlEndpoint;
-
-import com.hp.hpl.jena.query.ResultSet;
-
-public class AutomaticExampleFinderSPARQL {
-
- private static Logger logger = Logger
- .getLogger(ComponentManager.class);
-
- private Cache c;
- private SparqlEndpoint se;
- private SortedSet<String> posExamples;
- private SortedSet<String> negExamples;
-
-
- public AutomaticExampleFinderSPARQL(SparqlEndpoint se){
- this.c=new Cache();
- this.se=se;
- posExamples = new TreeSet<String>();
- negExamples = new TreeSet<String>();
- }
-
- public void init(String concept, boolean useRelated, boolean useSuperclasses,boolean useParallelClasses, int poslimit, int neglimit) {
- makePositiveExamplesFromConcept( concept);
- SortedSet<String> keepForClean = new TreeSet<String>();
- keepForClean.addAll(this.posExamples);
- this.posExamples = SetManipulation.fuzzyShrink(this.posExamples, poslimit);
- logger.trace("shrinking: pos Example size: "+posExamples.size());
-
- if(useRelated) {
- dbpediaMakeNegativeExamplesFromRelatedInstances(this.posExamples);
- }
- if(useSuperclasses) {
- dbpediaMakeNegativeExamplesFromSuperClasses(concept);
- }
- if(useParallelClasses) {
- dbpediaMakeNegativeExamplesFromClassesOfInstances();
- }
- //clean
- negExamples.removeAll(keepForClean);
- logger.trace("neg Example size after cleaning: "+negExamples.size());
- this.negExamples = SetManipulation.fuzzyShrink(negExamples, neglimit);
- logger.debug("pos Example size after shrinking: "+posExamples.size());
- logger.debug("neg Example size after shrinking: "+negExamples.size());
- logger.debug("Finished examples for concept: "+concept);
- }
-
- public SortedSet<String> getPosOnly(String concept, int limit){
- makePositiveExamplesFromConcept( concept);
- return SetManipulation.fuzzyShrink(this.posExamples, limit);
- }
-
- private void makePositiveExamplesFromConcept(String concept){
- logger.debug("making Positive Examples from Concept: "+concept);
- if(concept.contains("http://dbpedia.org/resource/Category:")) {
- this.posExamples = new JenaResultSetConvenience(querySKOSConcept(concept,0))
- .getStringListForVariable("subject");
- }else {
- this.posExamples = new JenaResultSetConvenience(queryConcept(concept,0))
- .getStringListForVariable("subject");
- }
- logger.debug(" pos Example size: "+posExamples.size());
- }
-
-
-
-
-
- private void dbpediaMakeNegativeExamplesFromRelatedInstances(SortedSet<String> subject) {
- logger.debug("making examples from related instances");
- for (String string : subject) {
- dbpediaMakeNegativeExamplesFromRelatedInstances(string);
- }
- logger.debug(" negExample size: "+negExamples.size());
- }
-
-
- /**
- *
- * @param subject
- * @return
- */
- private void dbpediaMakeNegativeExamplesFromRelatedInstances(String subject) {
- // SortedSet<String> result = new TreeSet<String>();
-
- String query = "SELECT * WHERE { \n" + "<" + subject + "> " + "?p ?o. \n"
- + "FILTER (REGEX(str(?o), 'http://dbpedia.org/resource/')).\n"
- + "FILTER (!REGEX(str(?p), 'http://www.w3.org/2004/02/skos'))\n"
- + "}";
-
- String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
- ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
- JenaResultSetConvenience rsc = new JenaResultSetConvenience(rs);
- this.negExamples.addAll(rsc.getStringListForVariable("o"));
-
-
- }
-
-
- private void dbpediaMakeNegativeExamplesFromSuperClasses(String concept) {
-
- SortedSet<String> superClasses = new TreeSet<String>();
- superClasses.add(concept.replace("\"", ""));
- //logger.debug("before"+superClasses);
- superClasses = dbpediaGetSuperClasses( superClasses, 4);
- logger.debug("making neg Examples from "+superClasses.size()+" superclasses");
- JenaResultSetConvenience rsc;
- for (String oneSuperClass : superClasses) {
-
- rsc = new JenaResultSetConvenience(queryConcept("\""+oneSuperClass+"\"", 0));
- this.negExamples.addAll(rsc.getStringListForVariable("subject"));
- }
- logger.debug(" neg Example size: "+negExamples.size());
- }
-
- private void dbpediaMakeNegativeExamplesFromClassesOfInstances() {
- logger.debug("making neg Examples from parallel classes");
- SortedSet<String> classes = new TreeSet<String>();
- //superClasses.add(concept.replace("\"", ""));
- //logger.debug("before"+superClasses);
- //superClasses = dbpediaGetSuperClasses( superClasses, 4);
- //logger.debug("getting negExamples from "+superClasses.size()+" superclasses");
- JenaResultSetConvenience rsc;
- ResultSet rs=null;
- for (String instance : posExamples) {
- //System.out.println(instance);
- rs = getClassesForInstance(instance);
- //System.out.println(ResultSetFormatter.asXMLString(rs));
- rsc = new JenaResultSetConvenience(rs);
- classes.addAll(rsc.getStringListForVariable("subject"));
- //System.out.println(classes);
- }
- logger.debug("getting negExamples from "+classes.size()+" parallel classes");
- for (String oneClass : classes) {
-
- rsc = new JenaResultSetConvenience(queryConcept("\""+oneClass+"\"",0));
- this.negExamples.addAll(rsc.getStringListForVariable("subject"));
- }
- logger.debug("neg Example size: "+negExamples.size());
-
- }
-
- private SortedSet<String> dbpediaGetSuperClasses(SortedSet<String> superClasses, int depth) {
- SortedSet<String> ret = new TreeSet<String>();
- SortedSet<String> tmpset = new TreeSet<String>();
- ret.addAll(superClasses);
- //logger.debug(superClasses);
- JenaResultSetConvenience rsc;
-
- String query = "";
- for (; depth != 0 ; depth--) {
- for (String oneSuperClass : superClasses) {
- //logger.debug("one"+oneSuperClass);
- //tmp = oneSuperClass.replace("\"", "");
- query = "SELECT * WHERE { \n" + "<" + oneSuperClass + "> "
- + "<http://www.w3.org/2000/01/rdf-schema#subClassOf> ?superclass. \n"
- + "}";
- String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
- ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
- rsc = new JenaResultSetConvenience(rs);
- tmpset.addAll(rsc.getStringListForVariable("superclass"));
- }
- ret.addAll(tmpset);
- //logger.debug(ret);
- superClasses.clear();
- superClasses.addAll(tmpset);
- tmpset.clear();
- }
- //logger.debug(concept);
- //logger.debug(query);
- return ret;
- }
-
- public ResultSet queryConcept(String concept,int limit) {
- ResultSet rs = null;
- try {
- String query = SparqlQueryDescriptionConvertVisitor
- .getSparqlQuery(concept,limit);
-
- SparqlQuery sq = new SparqlQuery(query, se);
- String JSON = c.executeSparqlQuery(sq);
- //System.out.println("JSON:\n"+JSON);
- rs = SparqlQuery.JSONtoResultSet(JSON);
-
- } catch (Exception e) {
- e.printStackTrace();
- }
-
- return rs;
- }
-
- public ResultSet querySKOSConcept(String SKOSconcept,int limit) {
- if(limit==0)limit=99999;
- //
- ResultSet rs = null;
- try {
-
- String query = "SELECT * WHERE { \n " +
- "?subject " +
- "<http://www.w3.org/2004/02/skos/core#subject> " +
- "<" + SKOSconcept + "> \n" +
- "} LIMIT "+limit;
- SparqlQuery sq = new SparqlQuery(query, se);
- String JSON = c.executeSparqlQuery(sq);
- //System.out.println(JSON);
- rs = SparqlQuery.JSONtoResultSet(JSON);
-
- } catch (Exception e) {
- e.printStackTrace();
- }
-
- return rs;
- }
-
- public ResultSet getClassesForInstance(String instance) {
- ResultSet rs = null;
- try {
-
- String query = "SELECT ?subject WHERE { \n " +
- "<" + instance + ">"+
- " a " +
- "?subject " +
- "\n" +
- "}";
- SparqlQuery sq = new SparqlQuery(query, se);
- //System.out.println(query);
- String JSON = c.executeSparqlQuery(sq);
- //System.out.println(JSON);
- rs = SparqlQuery.JSONtoResultSet(JSON);
-
- } catch (Exception e) {
- e.printStackTrace();
- }
-
- return rs;
- }
-
- public SortedSet<String> getPosExamples() {
- return posExamples;
- }
-
- public SortedSet<String> getNegExamples() {
- return negExamples;
- }
-
-
- /**
- * NOT WORKING
- * @param description
- */
- public void getSubClasses(String description) {
- ResultSet rs = null;
- try {
- String query = SparqlQueryDescriptionConvertVisitor
- .getSparqlSubclassQuery(description.replace("\"", ""));
-
- rs = new SparqlQuery(query, se).send();
- System.out.println(query);
- //System.out.println(SparqlQuery.getAsXMLString(rs));
- System.out.println(rs.getResultVars());
- SortedSet<String> remainingClasses = new JenaResultSetConvenience(rs).getStringListForVariable("subject");
- SortedSet<String> alreadyQueried = new TreeSet<String>();
- alreadyQueried.add(description);
- while (remainingClasses.size()!=0){
- String tmp = remainingClasses.first();
- remainingClasses.remove(tmp);
- query = SparqlQueryDescriptionConvertVisitor
- .getSparqlSubclassQuery(tmp);
- alreadyQueried.add(tmp);
- rs = new SparqlQuery(query, se).send();
- remainingClasses.addAll(new JenaResultSetConvenience(rs).getStringListForVariable("subject"));
- }
- //System.out.println(JSON);
- System.out.println(alreadyQueried);
-
- } catch (Exception e) {
- e.printStackTrace();
- }
-
- }
-
-
-
-}
Added: trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java (rev 0)
+++ trunk/src/dl-learner/org/dllearner/utilities/AutomaticExampleFinderSPARQL.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -0,0 +1,374 @@
+package org.dllearner.utilities;
+
+import java.util.SortedSet;
+import java.util.TreeSet;
+
+import org.apache.log4j.Logger;
+import org.dllearner.core.ComponentManager;
+import org.dllearner.kb.sparql.Cache;
+import org.dllearner.kb.sparql.SparqlQuery;
+import org.dllearner.kb.sparql.SparqlQueryDescriptionConvertVisitor;
+import org.dllearner.kb.sparql.configuration.SparqlEndpoint;
+
+import com.hp.hpl.jena.query.ResultSet;
+
+public class AutomaticExampleFinderSPARQL {
+
+ private static Logger logger = Logger
+ .getLogger(ComponentManager.class);
+
+ private Cache c;
+ private SparqlEndpoint se;
+ private SortedSet<String> posExamples;
+ private SortedSet<String> negExamples;
+
+
+ public AutomaticExampleFinderSPARQL(SparqlEndpoint se){
+ this.c=new Cache();
+ this.se=se;
+ posExamples = new TreeSet<String>();
+ negExamples = new TreeSet<String>();
+ }
+
+ public void initDBpedia(String concept, boolean useRelated, boolean useSuperclasses,boolean useParallelClasses, int poslimit, int neglimit) {
+ dbpediaMakePositiveExamplesFromConcept( concept);
+ SortedSet<String> keepForClean = new TreeSet<String>();
+ keepForClean.addAll(this.posExamples);
+ this.posExamples = SetManipulation.fuzzyShrink(this.posExamples, poslimit);
+ logger.trace("shrinking: pos Example size: "+posExamples.size());
+
+ if(useRelated) {
+ dbpediaMakeNegativeExamplesFromRelatedInstances(this.posExamples);
+ }
+ if(useSuperclasses) {
+ makeNegativeExamplesFromSuperClasses(concept);
+ }
+ if(useParallelClasses) {
+ makeNegativeExamplesFromClassesOfInstances();
+ }
+ //clean
+ negExamples.removeAll(keepForClean);
+ logger.trace("neg Example size after cleaning: "+negExamples.size());
+ this.negExamples = SetManipulation.fuzzyShrink(negExamples, neglimit);
+ logger.debug("pos Example size after shrinking: "+posExamples.size());
+ logger.debug("neg Example size after shrinking: "+negExamples.size());
+ logger.debug("Finished examples for concept: "+concept);
+ }
+
+ public void init(String concept, String namespace, boolean useRelated, boolean useSuperclasses,boolean useParallelClasses, int poslimit, int neglimit) {
+ makePositiveExamplesFromConcept( concept);
+ SortedSet<String> keepForClean = new TreeSet<String>();
+ keepForClean.addAll(this.posExamples);
+ this.posExamples = SetManipulation.fuzzyShrink(this.posExamples, poslimit);
+ logger.trace("shrinking: pos Example size: "+posExamples.size());
+
+ if(useRelated) {
+ makeNegativeExamplesFromRelatedInstances(this.posExamples,namespace);
+ }
+ if(useSuperclasses) {
+ makeNegativeExamplesFromSuperClasses(concept);
+ }
+ if(useParallelClasses) {
+ makeNegativeExamplesFromClassesOfInstances();
+ }
+ //clean
+ negExamples.removeAll(keepForClean);
+ logger.trace("neg Example size after cleaning: "+negExamples.size());
+ this.negExamples = SetManipulation.fuzzyShrink(negExamples, neglimit);
+ logger.debug("pos Example size after shrinking: "+posExamples.size());
+ logger.debug("neg Example size after shrinking: "+negExamples.size());
+ logger.debug("Finished examples for concept: "+concept);
+ }
+
+
+
+ public SortedSet<String> dbpediaGetPosOnly(String concept, int limit){
+ dbpediaMakePositiveExamplesFromConcept( concept);
+ return SetManipulation.fuzzyShrink(this.posExamples, limit);
+ }
+
+ public SortedSet<String> getPosOnly(String concept, int limit){
+ makePositiveExamplesFromConcept( concept);
+ return SetManipulation.fuzzyShrink(this.posExamples, limit);
+ }
+
+ private void dbpediaMakePositiveExamplesFromConcept(String concept){
+ logger.debug("making Positive Examples from Concept: "+concept);
+ if(concept.contains("http://dbpedia.org/resource/Category:")) {
+ this.posExamples = new JenaResultSetConvenience(dbpediaQuerySKOSConcept(concept,0))
+ .getStringListForVariable("subject");
+ }else {
+ this.posExamples = new JenaResultSetConvenience(queryConcept(concept,0))
+ .getStringListForVariable("subject");
+ }
+ logger.debug(" pos Example size: "+posExamples.size());
+ }
+
+ private void makePositiveExamplesFromConcept(String concept){
+ logger.debug("making Positive Examples from Concept: "+concept);
+ this.posExamples = new JenaResultSetConvenience(queryConcept(concept,0))
+ .getStringListForVariable("subject");
+ logger.debug(" pos Example size: "+posExamples.size());
+ }
+
+
+
+ /*private void makePositiveExamplesFromConcept(String concept){
+ logger.debug("making Positive Examples from Concept: "+concept);
+ if(concept.contains("http://dbpedia.org/resource/Category:")) {
+ this.posExamples = new JenaResultSetConvenience(querySKOSConcept(concept,0))
+ .getStringListForVariable("subject");
+ }else {
+ this.posExamples = new JenaResultSetConvenience(queryConcept(concept,0))
+ .getStringListForVariable("subject");
+ }
+ logger.debug(" pos Example size: "+posExamples.size());
+ }*/
+
+
+
+
+
+ private void dbpediaMakeNegativeExamplesFromRelatedInstances(SortedSet<String> subject) {
+ logger.debug("making examples from related instances");
+ for (String string : subject) {
+ dbpediaMakeNegativeExamplesFromRelatedInstances(string);
+ }
+ logger.debug(" negExample size: "+negExamples.size());
+ }
+
+ private void makeNegativeExamplesFromRelatedInstances(SortedSet<String> subject, String namespace) {
+ logger.debug("making examples from related instances");
+ for (String string : subject) {
+ makeNegativeExamplesFromRelatedInstances(string,namespace);
+ }
+ logger.debug(" negExample size: "+negExamples.size());
+ }
+
+
+ /**
+ *
+ * @param subject
+ * @return
+ */
+ private void dbpediaMakeNegativeExamplesFromRelatedInstances(String subject) {
+ // SortedSet<String> result = new TreeSet<String>();
+
+ String query = "SELECT * WHERE { \n" + "<" + subject + "> " + "?p ?o. \n"
+ + "FILTER (REGEX(str(?o), 'http://dbpedia.org/resource/')).\n"
+ + "FILTER (!REGEX(str(?p), 'http://www.w3.org/2004/02/skos'))\n"
+ + "}";
+
+ String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
+ ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
+ JenaResultSetConvenience rsc = new JenaResultSetConvenience(rs);
+ this.negExamples.addAll(rsc.getStringListForVariable("o"));
+
+
+ }
+
+ private void makeNegativeExamplesFromRelatedInstances(String subject, String namespace) {
+ // SortedSet<String> result = new TreeSet<String>();
+
+ String query = "SELECT * WHERE { \n" + "<" + subject + "> " + "?p ?o. \n"
+ + "FILTER (REGEX(str(?o), '"+namespace+"')).\n"
+ + "}";
+
+ String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
+ ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
+ JenaResultSetConvenience rsc = new JenaResultSetConvenience(rs);
+ this.negExamples.addAll(rsc.getStringListForVariable("o"));
+
+
+ }
+
+
+ private void makeNegativeExamplesFromSuperClasses(String concept) {
+
+ SortedSet<String> superClasses = new TreeSet<String>();
+ superClasses.add(concept.replace("\"", ""));
+ //logger.debug("before"+superClasses);
+ superClasses = getSuperClasses( superClasses, 4);
+ logger.debug("making neg Examples from "+superClasses.size()+" superclasses");
+ JenaResultSetConvenience rsc;
+ for (String oneSuperClass : superClasses) {
+
+ rsc = new JenaResultSetConvenience(queryConcept("\""+oneSuperClass+"\"", 0));
+ this.negExamples.addAll(rsc.getStringListForVariable("subject"));
+ }
+ logger.debug(" neg Example size: "+negExamples.size());
+ }
+
+
+
+ private void makeNegativeExamplesFromClassesOfInstances() {
+ logger.debug("making neg Examples from parallel classes");
+ SortedSet<String> classes = new TreeSet<String>();
+ //superClasses.add(concept.replace("\"", ""));
+ //logger.debug("before"+superClasses);
+ //superClasses = dbpediaGetSuperClasses( superClasses, 4);
+ //logger.debug("getting negExamples from "+superClasses.size()+" superclasses");
+ JenaResultSetConvenience rsc;
+ ResultSet rs=null;
+ for (String instance : posExamples) {
+ //System.out.println(instance);
+ rs = getClassesForInstance(instance);
+ //System.out.println(ResultSetFormatter.asXMLString(rs));
+ rsc = new JenaResultSetConvenience(rs);
+ classes.addAll(rsc.getStringListForVariable("subject"));
+ //System.out.println(classes);
+ }
+ logger.debug("getting negExamples from "+classes.size()+" parallel classes");
+ for (String oneClass : classes) {
+
+ rsc = new JenaResultSetConvenience(queryConcept("\""+oneClass+"\"",0));
+ this.negExamples.addAll(rsc.getStringListForVariable("subject"));
+ }
+ logger.debug("neg Example size: "+negExamples.size());
+
+ }
+
+ private SortedSet<String> getSuperClasses(SortedSet<String> superClasses, int depth) {
+ SortedSet<String> ret = new TreeSet<String>();
+ SortedSet<String> tmpset = new TreeSet<String>();
+ ret.addAll(superClasses);
+ //logger.debug(superClasses);
+ JenaResultSetConvenience rsc;
+
+ String query = "";
+ for (; depth != 0 ; depth--) {
+ for (String oneSuperClass : superClasses) {
+ //logger.debug("one"+oneSuperClass);
+ //tmp = oneSuperClass.replace("\"", "");
+ query = "SELECT * WHERE { \n" + "<" + oneSuperClass + "> "
+ + "<http://www.w3.org/2000/01/rdf-schema#subClassOf> ?superclass. \n"
+ + "}";
+ String JSON = (c.executeSparqlQuery(new SparqlQuery(query, se)));
+ ResultSet rs =SparqlQuery.JSONtoResultSet(JSON);
+ rsc = new JenaResultSetConvenience(rs);
+ tmpset.addAll(rsc.getStringListForVariable("superclass"));
+ }
+ ret.addAll(tmpset);
+ //logger.debug(ret);
+ superClasses.clear();
+ superClasses.addAll(tmpset);
+ tmpset.clear();
+ }
+ //logger.debug(concept);
+ //logger.debug(query);
+ return ret;
+ }
+
+
+
+ public ResultSet queryConcept(String concept,int limit) {
+ ResultSet rs = null;
+ try {
+ String query = SparqlQueryDescriptionConvertVisitor
+ .getSparqlQuery(concept,limit);
+
+ SparqlQuery sq = new SparqlQuery(query, se);
+ String JSON = c.executeSparqlQuery(sq);
+ //System.out.println("JSON:\n"+JSON);
+ rs = SparqlQuery.JSONtoResultSet(JSON);
+
+ } catch (Exception e) {
+ e.printStackTrace();
+ }
+
+ return rs;
+ }
+
+ public ResultSet dbpediaQuerySKOSConcept(String SKOSconcept,int limit) {
+ if(limit==0)limit=99999;
+ //
+ ResultSet rs = null;
+ try {
+
+ String query = "SELECT * WHERE { \n " +
+ "?subject " +
+ "<http://www.w3.org/2004/02/skos/core#subject> " +
+ "<" + SKOSconcept + "> \n" +
+ "} LIMIT "+limit;
+ SparqlQuery sq = new SparqlQuery(query, se);
+ String JSON = c.executeSparqlQuery(sq);
+ //System.out.println(JSON);
+ rs = SparqlQuery.JSONtoResultSet(JSON);
+
+ } catch (Exception e) {
+ e.printStackTrace();
+ }
+
+ return rs;
+ }
+
+ public ResultSet getClassesForInstance(String instance) {
+ ResultSet rs = null;
+ try {
+
+ String query = "SELECT ?subject WHERE { \n " +
+ "<" + instance + ">"+
+ " a " +
+ "?subject " +
+ "\n" +
+ "}";
+ SparqlQuery sq = new SparqlQuery(query, se);
+ //System.out.println(query);
+ String JSON = c.executeSparqlQuery(sq);
+ //System.out.println(JSON);
+ rs = SparqlQuery.JSONtoResultSet(JSON);
+
+ } catch (Exception e) {
+ e.printStackTrace();
+ }
+
+ return rs;
+ }
+
+ public SortedSet<String> getPosExamples() {
+ return posExamples;
+ }
+
+ public SortedSet<String> getNegExamples() {
+ return negExamples;
+ }
+
+
+ /**
+ * NOT WORKING
+ * @param description
+ */
+ public void getSubClasses(String description) {
+ ResultSet rs = null;
+ try {
+ String query = SparqlQueryDescriptionConvertVisitor
+ .getSparqlSubclassQuery(description.replace("\"", ""));
+
+ rs = new SparqlQuery(query, se).send();
+ System.out.println(query);
+ //System.out.println(SparqlQuery.getAsXMLString(rs));
+ System.out.println(rs.getResultVars());
+ SortedSet<String> remainingClasses = new JenaResultSetConvenience(rs).getStringListForVariable("subject");
+ SortedSet<String> alreadyQueried = new TreeSet<String>();
+ alreadyQueried.add(description);
+ while (remainingClasses.size()!=0){
+ String tmp = remainingClasses.first();
+ remainingClasses.remove(tmp);
+ query = SparqlQueryDescriptionConvertVisitor
+ .getSparqlSubclassQuery(tmp);
+ alreadyQueried.add(tmp);
+ rs = new SparqlQuery(query, se).send();
+ remainingClasses.addAll(new JenaResultSetConvenience(rs).getStringListForVariable("subject"));
+ }
+ //System.out.println(JSON);
+ System.out.println(alreadyQueried);
+
+ } catch (Exception e) {
+ e.printStackTrace();
+ }
+
+ }
+
+
+
+}
Modified: trunk/src/dl-learner/org/dllearner/utilities/ConfWriter.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/utilities/ConfWriter.java 2008-05-06 15:02:50 UTC (rev 829)
+++ trunk/src/dl-learner/org/dllearner/utilities/ConfWriter.java 2008-05-06 18:16:26 UTC (rev 830)
@@ -43,23 +43,12 @@
}
public void writeSPARQL(String filename, SortedSet<String> pos, SortedSet<String> neg,
- String uri, SortedSet<String> ignoredConcepts) {
+ String uri, SortedSet<String> ignoredConcepts, String standardSettings, String prefixAlgortihm) {
- String prefixAlgortihm = "refexamples";
+
String prefixSparql = "sparql";
- String standardSettings =
- "sparql.recursionDepth = 1;\n" +
- "sparql.predefinedFilter = \"YAGO\";\n" +
- "sparql.predefinedEndpoint = \"DBPEDIA\";\n" +
- "refexamples.minExecutionTimeInSeconds = 30;\n" +
- "refexamples.maxExecutionTimeInSeconds = 30;\n" +
- "//refexamples.guaranteeXgoodDescriptions = 10;\n" +
- "refexamples.logLevel=\"TRACE\";\n" +
- "refexamples.noisePercentage = 0.10;\n" +
- "refexamples.writeSearchTree = true;\n" +
- "refexamples.searchTreeFile = \"searchTree.txt\";\n"+
- "refexamples.replaceSearchTree = true;\n\n" ;
+
// "sparql.format = \"KB\";\n\n";
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
