Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3835] trunk/components-ext/src
|
From: <ki...@us...> - 2012-08-23 14:35:44
|
Revision: 3835
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3835&view=rev
Author: kirdie
Date: 2012-08-23 14:35:33 +0000 (Thu, 23 Aug 2012)
Log Message:
-----------
more work on the oxford test.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner2.java
trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/QueryTestData.java
trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner3Test.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner2.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner2.java 2012-08-23 13:19:03 UTC (rev 3834)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner2.java 2012-08-23 14:35:33 UTC (rev 3835)
@@ -911,6 +911,7 @@
// return 0;
// }
// return Math.log(cnt);
+ if(Double.isNaN(popularity)) {throw new AssertionError("prominence NaN for uri "+uri+", slot type "+type);}
return popularity;
}
Modified: trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/QueryTestData.java
===================================================================
--- trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/QueryTestData.java 2012-08-23 13:19:03 UTC (rev 3834)
+++ trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/QueryTestData.java 2012-08-23 14:35:33 UTC (rev 3835)
@@ -33,6 +33,7 @@
import org.w3c.dom.Element;
import org.w3c.dom.NodeList;
import org.xml.sax.SAXException;
+import com.hp.hpl.jena.rdf.model.Model;
class QueryTestData implements Serializable
{
@@ -68,12 +69,12 @@
catch (ClassNotFoundException e){throw new RuntimeException(e);}
}
- public QueryTestData generateAnswers(SparqlEndpoint endpoint, ExtractionDBCache cache)
+ public QueryTestData generateAnswers(SparqlEndpoint endpoint, ExtractionDBCache cache,Model model)
{
if(!id2Answers.isEmpty()) {throw new AssertionError("Answers already existing.");}
for(int i:id2Query.keySet())
{
- Set<String> uris = SPARQLTemplateBasedLearner3Test.getUris(endpoint, id2Query.get(i),cache);
+ Set<String> uris = SPARQLTemplateBasedLearner3Test.getUris(endpoint, id2Query.get(i),cache,model);
id2Answers.put(i, uris); // empty answer set better transfers intended meaning and doesn't cause NPEs in html generation :-)
if(!uris.isEmpty()) {/*id2Answers.put(i, uris);*/}
else {id2LearnStatus.put(i, LearnStatus.QUERY_RESULT_EMPTY);}
Modified: trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner3Test.java
===================================================================
--- trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner3Test.java 2012-08-23 13:19:03 UTC (rev 3834)
+++ trunk/components-ext/src/test/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner3Test.java 2012-08-23 14:35:33 UTC (rev 3835)
@@ -50,6 +50,7 @@
import org.apache.log4j.SimpleLayout;
import org.dllearner.algorithm.tbsl.ltag.parser.Parser;
import org.dllearner.algorithm.tbsl.nlp.PartOfSpeechTagger;
+import org.dllearner.algorithm.tbsl.nlp.StanfordPartOfSpeechTagger;
import org.dllearner.algorithm.tbsl.nlp.SynchronizedStanfordPartOfSpeechTagger;
import org.dllearner.algorithm.tbsl.nlp.WordNet;
import org.dllearner.algorithm.tbsl.templator.Templator;
@@ -70,8 +71,11 @@
import org.w3c.dom.NodeList;
import org.xml.sax.SAXException;
import cern.colt.Arrays;
+import com.hp.hpl.jena.query.QueryExecutionFactory;
+import com.hp.hpl.jena.query.QueryFactory;
import com.hp.hpl.jena.query.QuerySolution;
import com.hp.hpl.jena.query.ResultSet;
+import com.hp.hpl.jena.query.Syntax;
import com.hp.hpl.jena.rdf.model.Model;
import com.hp.hpl.jena.rdf.model.ModelFactory;
import com.hp.hpl.jena.rdf.model.RDFNode;
@@ -104,17 +108,25 @@
{test("QALD 2 Benchmark ideally tagged", new File(getClass().getClassLoader().getResource("tbsl/evaluation/qald2-dbpedia-train-tagged(ideal).xml").getFile()),
SparqlEndpoint.getEndpointDBpedia(),dbpediaLiveCache);}
+ /*@Test*/ public void testOxford() throws Exception
+ {
+ }
+
@Test public void generateXMLOxford() throws IOException
{
- Model m = loadOxfordModel();
+ boolean ADD_POS_TAGS = false;
+ PartOfSpeechTagger posTagger = new StanfordPartOfSpeechTagger();
+ Model model = loadOxfordModel();
List<String> questions = new LinkedList<String>();
BufferedReader in = new BufferedReader((new InputStreamReader(getClass().getClassLoader().getResourceAsStream("tbsl/oxford_eval_queries.txt"))));
int j=0;
for(String line;(line=in.readLine())!=null;)
{
j++;
- if(j>1) break;
- if(!line.isEmpty()) {questions.add(line.replace("question: ", ""));}
+ // if(j>5) break; // TODO: remove later
+ String question = line.replace("question: ", "").trim();
+ if(ADD_POS_TAGS&&!OXFORD_PRETAGGED) {question = posTagger.tag(question);}
+ if(!line.trim().isEmpty()) {questions.add(question);}
}
in.close();
SortedMap<Integer,String> id2Question = new TreeMap<Integer, String>();
@@ -126,7 +138,8 @@
SPARQLTemplateBasedLearner2.class.getClassLoader().getResource("tbsl/oxford_dataproperty_mappings.txt").getPath(),
SPARQLTemplateBasedLearner2.class.getClassLoader().getResource("tbsl/oxford_objectproperty_mappings.txt").getPath()
);
- QueryTestData testData = generateTestDataMultiThreaded(id2Question, null,m,mappingIndex,OXFORD_PRETAGGED);
+ QueryTestData testData = generateTestDataMultiThreaded(id2Question, null,model,mappingIndex,ADD_POS_TAGS||OXFORD_PRETAGGED);
+ testData.generateAnswers(null, null, model);
testData.writeQaldXml(new File("log/test.xml"));
}
@@ -175,7 +188,7 @@
String type = (ending.equals("ttl")||ending.equals("nt"))?"TURTLE":ending.equals("owl")?"RDF/XML":String.valueOf(Integer.valueOf("filetype "+ending+" not handled."));
// switch(type) {case "ttl":type="TURTLE";break;case "owl":type="RDF/XML";break;default:throw new RuntimeException("filetype "+ending+" not handled.");} // no Java 1.7 :-(
try{
-// m.read(new FileInputStream(new File("/home/lorenz/arbeit/papers/question-answering-iswc-2012/data/"+s)), null, type);}catch (FileNotFoundException e) {}
+ // m.read(new FileInputStream(new File("/home/lorenz/arbeit/papers/question-answering-iswc-2012/data/"+s)), null, type);}catch (FileNotFoundException e) {}
m.read(getClass().getClassLoader().getResourceAsStream("oxford/"+s),null, type);}
catch(RuntimeException e) {throw new RuntimeException("Could not read into model: "+s,e);}
}
@@ -241,7 +254,7 @@
if(!updatedReferenceXML.exists())
{
logger.info("Generating updated reference for "+title);
- generateUpdatedXML(referenceXML,updatedReferenceXML,endpoint,cache);
+ generateUpdatedXML(referenceXML,updatedReferenceXML,endpoint,cache,null);
}
QueryTestData referenceTestData = QueryTestData.readQaldXml(updatedReferenceXML);
@@ -251,7 +264,7 @@
QueryTestData learnedTestData = generateTestDataMultiThreaded(referenceTestData.id2Question, dbpediaLiveKnowledgebase,null,null,DBPEDIA_PRETAGGED);
long endLearning = System.currentTimeMillis();
logger.info("finished learning after "+(endLearning-startLearning)/1000.0+"s");
- learnedTestData.generateAnswers(endpoint,cache);
+ learnedTestData.generateAnswers(endpoint,cache,null);
long endGeneratingAnswers = System.currentTimeMillis();
logger.info("finished generating answers in "+(endGeneratingAnswers-endLearning)/1000.0+"s");
Evaluation evaluation = evaluate(referenceTestData, learnedTestData);
@@ -259,6 +272,11 @@
evaluation.write();
}
+ private void evaluateAndWrite()
+ {
+
+ }
+
/** evaluates a data set against a reference.
* @param reference the test data assumed to be correct. needs to contain the answers for all queries.
* @param suspect the test data to compare with the reference.
@@ -542,7 +560,8 @@
}
catch (ExecutionException e)
{
- throw new RuntimeException("question="+question,e);
+ testData.id2LearnStatus.put(i, new LearnStatus(LearnStatus.Type.EXCEPTION, e));
+ //throw new RuntimeException("question="+question,e);
}
catch (TimeoutException e)
{
@@ -590,7 +609,7 @@
* @throws SAXException
* @throws TransformerException
*/
- private void generateUpdatedXML(File originalFile, File updatedFile,SparqlEndpoint endpoint, ExtractionDBCache cache) throws ParserConfigurationException, SAXException, IOException, TransformerException
+ private void generateUpdatedXML(File originalFile, File updatedFile,SparqlEndpoint endpoint, ExtractionDBCache cache,Model model) throws ParserConfigurationException, SAXException, IOException, TransformerException
{
logger.info(String.format("Updating question file \"%s\" by removing questions without nonempty resource list answer and adding answers.\n" +
" Saving the result to file \"%s\"",originalFile.getPath(),updatedFile.getPath()));
@@ -629,7 +648,7 @@
if(!query.equals("OUT OF SCOPE")) // marker in qald benchmark file, will create holes interval of ids (e.g. 1,2,5,7)
{
- Set<String> uris = getUris(endpoint, query,cache);
+ Set<String> uris = getUris(endpoint, query,cache,model);
if(!uris.isEmpty())
{
// remove reference answers of the benchmark because they are obtained from an other endpoint
@@ -687,7 +706,7 @@
// private SPARQLTemplateBasedLearner2 oxfordLearner;
// private SPARQLTemplateBasedLearner2 dbpediaLiveLearner;
-// private final ExtractionDBCache oxfordCache = new ExtractionDBCache("cache");
+ // private final ExtractionDBCache oxfordCache = new ExtractionDBCache("cache");
private final static ExtractionDBCache dbpediaLiveCache = new ExtractionDBCache("cache");
private final Knowledgebase dbpediaLiveKnowledgebase = createDBpediaLiveKnowledgebase(dbpediaLiveCache);
@@ -725,7 +744,7 @@
Logger.getRootLogger().setLevel(Level.WARN);
Logger.getLogger(Templator.class).setLevel(Level.WARN);
Logger.getLogger(Parser.class).setLevel(Level.WARN);
- Logger.getLogger(SPARQLTemplateBasedLearner2.class).setLevel(Level.INFO);
+ Logger.getLogger(SPARQLTemplateBasedLearner2.class).setLevel(Level.WARN);
// Logger.getLogger(SPARQLTemplateBasedLearner2.class).setLevel(Level.INFO);
logger.setLevel(Level.INFO); // TODO: remove when finishing implementation of this class
logger.addAppender(new FileAppender(new SimpleLayout(), "log/"+this.getClass().getSimpleName()+".log", false));
@@ -734,16 +753,21 @@
// oxfordLearner = new SPARQLTemplateBasedLearner2(createOxfordKnowledgebase(oxfordCache));
}
- public static Set<String> getUris(final SparqlEndpoint endpoint, final String query, ExtractionDBCache cache)
+ public static Set<String> getUris(final SparqlEndpoint endpoint, final String query, ExtractionDBCache cache, Model model)
{
if(query==null) {throw new AssertionError("query is null");}
- if(endpoint==null) {throw new AssertionError("endpoint is null");}
+// if(endpoint==null) {throw new AssertionError("endpoint is null");}
if(!query.contains("SELECT")&&!query.contains("select")) {return Collections.<String>emptySet();} // abort when not a select query
Set<String> uris = new HashSet<String>();
// QueryEngineHTTP qe = new QueryEngineHTTP(DBPEDIA_LIVE_ENDPOINT_URL_STRING, query);
+
ResultSet rs;
// try{rs = qe.execSelect();}
- try{rs = executeSelect(endpoint, query, cache);}
+ try
+ {
+ if(model!=null) {rs = QueryExecutionFactory.create(QueryFactory.create(query, Syntax.syntaxARQ), model).execSelect();}
+ else {rs = executeSelect(endpoint, query, cache);}
+ }
catch(QueryExceptionHTTP e)
{
logger.error("Error getting uris for query "+query+" at endpoint "+endpoint,e);
@@ -780,7 +804,7 @@
}
return uris;
}
-
+
private static String urlDecode(String url){
String decodedURL = null;
try {
@@ -836,7 +860,7 @@
static private class POSTaggerHolder
{static public final PartOfSpeechTagger posTagger = new SynchronizedStanfordPartOfSpeechTagger();}
-
+
static private final WordNet wordnet = new WordNet();
static private final Options options = new Options();
private final SPARQLTemplateBasedLearner2 learner;
@@ -848,6 +872,7 @@
this.testData=testData;
learner = new SPARQLTemplateBasedLearner2(knowledgeBase,pretagged?null:POSTaggerHolder.posTagger,wordnet,options);
try {learner.init();} catch (ComponentInitException e) {throw new RuntimeException(e);}
+ learner.setUseIdealTagger(pretagged);
}
public LearnQueryCallable(String question, int id, QueryTestData testData, Model model,MappingBasedIndex index,boolean pretagged)
@@ -861,16 +886,18 @@
SPARQLTemplateBasedLearner2.class.getClassLoader().getResource("tbsl/oxford_dataproperty_mappings.txt").getPath(),
SPARQLTemplateBasedLearner2.class.getClassLoader().getResource("tbsl/oxford_objectproperty_mappings.txt").getPath()
);
-
+
learner = new SPARQLTemplateBasedLearner2(model,mappingIndex,pretagged?null:POSTaggerHolder.posTagger);
try {learner.init();} catch (ComponentInitException e) {throw new RuntimeException(e);}
+ learner.setUseIdealTagger(pretagged);
learner.setGrammarFiles(new String[]{"tbsl/lexicon/english.lex","tbsl/lexicon/english_oxford.lex"});
learner.setUseDomainRangeRestriction(false);
}
@Override public LearnStatus call()
- {
+ {
+
logger.trace("learning question: "+question);
try
{
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
