Menu
▾
▴
dl-learner-svn
|
From: <lor...@us...> - 2011-06-14 20:10:37
|
Revision: 2875
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=2875&view=rev
Author: lorenz_b
Date: 2011-06-14 20:10:31 +0000 (Tue, 14 Jun 2011)
Log Message:
-----------
Updated methods to select appropriate cache/index for new SLOT types.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-06-14 19:26:51 UTC (rev 2874)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-06-14 20:10:31 UTC (rev 2875)
@@ -529,11 +529,12 @@
private SolrSearch getIndexBySlotType(Slot slot){
SolrSearch index = null;
- if(slot.getSlotType() == SlotType.CLASS){
+ SlotType type = slot.getSlotType();
+ if(type == SlotType.CLASS){
index = class_index;
- } else if(slot.getSlotType() == SlotType.PROPERTY){
+ } else if(type == SlotType.PROPERTY || type == SlotType.SYMPROPERTY){
index = property_index;
- } else if(slot.getSlotType() == SlotType.RESOURCE){
+ } else if(type == SlotType.RESOURCE || type == SlotType.UNSPEC){
index = resource_index;
}
return index;
@@ -541,11 +542,12 @@
private Map<String, List<String>> getCacheBySlotType(Slot slot){
Map<String, List<String>> cache = null;
- if(slot.getSlotType() == SlotType.CLASS){
+ SlotType type = slot.getSlotType();
+ if(type == SlotType.CLASS){
cache = classesURICache;
- } else if(slot.getSlotType() == SlotType.PROPERTY){
+ } else if(type == SlotType.PROPERTY || type == SlotType.SYMPROPERTY){
cache = propertiesURICache;
- } else if(slot.getSlotType() == SlotType.RESOURCE){
+ } else if(type == SlotType.RESOURCE || type == SlotType.UNSPEC){
cache = resourcesURICache;
}
return cache;
@@ -641,7 +643,6 @@
}
private List<String> getResultFromRemoteEndpoint(String query){
- System.out.println(query);
List<String> resources = new ArrayList<String>();
try {
ResultSet rs = SparqlQuery.convertJSONtoResultSet(cache.executeSelectQuery(endpoint, query + " LIMIT 10"));
@@ -680,7 +681,7 @@
// Logger.getLogger(DefaultHttpParams.class).setLevel(Level.OFF);
// Logger.getLogger(HttpClient.class).setLevel(Level.OFF);
// Logger.getLogger(HttpMethodBase.class).setLevel(Level.OFF);
- String question = "Who are the presidents of the United States?";
+ String question = "Give me all actors starring in Batman Begins.";
// String question = "Give me all films starring Brad Pitt";
SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();
SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://live.dbpedia.org/sparql"),
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-08-04 10:53:05
|
Revision: 2988
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=2988&view=rev
Author: lorenz_b
Date: 2011-08-04 10:52:59 +0000 (Thu, 04 Aug 2011)
Log Message:
-----------
Added needed method init() caused by refactoring of core module.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-08-04 10:04:56 UTC (rev 2987)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-08-04 10:52:59 UTC (rev 2988)
@@ -35,6 +35,7 @@
import org.dllearner.algorithm.tbsl.templator.Templator;
import org.dllearner.algorithm.tbsl.util.Prefixes;
import org.dllearner.algorithm.tbsl.util.Similarity;
+import org.dllearner.core.ComponentInitException;
import org.dllearner.core.Oracle;
import org.dllearner.core.SparqlQueryLearningAlgorithm;
import org.dllearner.kb.sparql.ExtractionDBCache;
@@ -768,6 +769,12 @@
}
}
+ @Override
+ public void init() throws ComponentInitException {
+ // TODO Auto-generated method stub
+
+ }
+
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-01 13:48:42
|
Revision: 3349
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3349&view=rev
Author: lorenz_b
Date: 2011-11-01 13:48:33 +0000 (Tue, 01 Nov 2011)
Log Message:
-----------
fixed problem.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 13:31:55 UTC (rev 3348)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 13:48:33 UTC (rev 3349)
@@ -407,6 +407,11 @@
q.setScore(q.getScore()/t.getSlots().size());
}
allQueries.addAll(queries);
+ List<Query> qList = new ArrayList<Query>();
+ for(WeightedQuery wQ : queries){
+ qList.add(wQ.getQuery());
+ }
+ template2Queries.put(t, qList);
}
return allQueries;
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-01 14:08:19
|
Revision: 3350
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3350&view=rev
Author: lorenz_b
Date: 2011-11-01 14:08:09 +0000 (Tue, 01 Nov 2011)
Log Message:
-----------
Added limit for max tested queries.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 13:48:33 UTC (rev 3349)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 14:08:09 UTC (rev 3350)
@@ -79,6 +79,8 @@
private int maxTestedQueriesPerTemplate = 50;
private int maxQueryExecutionTimeInSeconds;
+ private int maxTestedQueries = 200;
+
private SparqlEndpoint endpoint = SparqlEndpoint.getEndpointDBpediaLiveAKSW();
private ExtractionDBCache cache = new ExtractionDBCache("cache");
@@ -248,8 +250,12 @@
//get the weighted query candidates
Set<WeightedQuery> weightedQueries = getWeightedSPARQLQueries(templates);
sparqlQueryCandidates = new ArrayList<Query>();
+ int i = 0;
for(WeightedQuery wQ : weightedQueries){
sparqlQueryCandidates.add(wQ.getQuery());
+ if(i == maxTestedQueries){
+ break;
+ }
}
//test candidates
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-01 14:55:51
|
Revision: 3351
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3351&view=rev
Author: lorenz_b
Date: 2011-11-01 14:55:44 +0000 (Tue, 01 Nov 2011)
Log Message:
-----------
Fixed problem.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 14:08:09 UTC (rev 3350)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-01 14:55:44 UTC (rev 3351)
@@ -256,6 +256,7 @@
if(i == maxTestedQueries){
break;
}
+ i++;
}
//test candidates
@@ -386,28 +387,31 @@
Set<WeightedQuery> tmp = new HashSet<WeightedQuery>();
for(Slot slot : t.getSlots()){
- for(Allocation a : slot2Allocations.get(slot)){
- for(WeightedQuery query : queries){
- if(slot.getSlotType() == SlotType.SYMPROPERTY){
- Query reversedQuery = new Query(query.getQuery());
- reversedQuery.getTriplesWithVar(slot.getAnchor()).iterator().next().reverse();
- reversedQuery.replaceVarWithURI(slot.getAnchor(), a.getUri());
- WeightedQuery w = new WeightedQuery(reversedQuery);
+ if(!slot2Allocations.get(slot).isEmpty()){
+ for(Allocation a : slot2Allocations.get(slot)){
+ for(WeightedQuery query : queries){
+ if(slot.getSlotType() == SlotType.SYMPROPERTY){
+ Query reversedQuery = new Query(query.getQuery());
+ reversedQuery.getTriplesWithVar(slot.getAnchor()).iterator().next().reverse();
+ reversedQuery.replaceVarWithURI(slot.getAnchor(), a.getUri());
+ WeightedQuery w = new WeightedQuery(reversedQuery);
+ double newScore = query.getScore() + a.getScore();
+ w.setScore(newScore);
+ tmp.add(w);
+ }
+ Query q = new Query(query.getQuery());
+ q.replaceVarWithURI(slot.getAnchor(), a.getUri());
+ WeightedQuery w = new WeightedQuery(q);
double newScore = query.getScore() + a.getScore();
w.setScore(newScore);
tmp.add(w);
}
- Query q = new Query(query.getQuery());
- q.replaceVarWithURI(slot.getAnchor(), a.getUri());
- WeightedQuery w = new WeightedQuery(q);
- double newScore = query.getScore() + a.getScore();
- w.setScore(newScore);
- tmp.add(w);
}
+ queries.clear();
+ queries.addAll(tmp);
+ tmp.clear();
}
- queries.clear();
- queries.addAll(tmp);
- tmp.clear();
+
}
for(WeightedQuery q : queries){
q.setScore(q.getScore()/t.getSlots().size());
@@ -465,6 +469,7 @@
return cnt;
}
+
private Map<Template, Collection<? extends Query>> getSPARQLQueryCandidates(Set<Template> templates){
logger.info("Generating candidate SPARQL queries...");
mon.start();
@@ -926,10 +931,11 @@
// Logger.getLogger(DefaultHttpParams.class).setLevel(Level.OFF);
// Logger.getLogger(HttpClient.class).setLevel(Level.OFF);
// Logger.getLogger(HttpMethodBase.class).setLevel(Level.OFF);
- String question = "Give me all films produced by Hal Roach?";
+// String question = "In which programming language is GIMP written?";
+ String question = "Who/WP are/VBP the/DT presidents/NNS of/IN the/DT United/NNP States/NNPS";
// String question = "Give me all books written by authors influenced by Ernest Hemingway.";
- SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();
+ SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();learner.setUseIdealTagger(true);
// SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://greententacle.techfak.uni-bielefeld.de:5171/sparql"),
// Collections.<String>singletonList(""), Collections.<String>emptyList());
SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://greententacle.techfak.uni-bielefeld.de:5171/sparql"),
@@ -939,7 +945,7 @@
learner.learnSPARQLQueries();
System.out.println("Learned query:\n" + learner.getBestSPARQLQuery());
System.out.println("Lexical answer type is: " + learner.getTemplates().iterator().next().getLexicalAnswerType());
-
+
}
@Override
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-04 15:03:18
|
Revision: 3373
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3373&view=rev
Author: lorenz_b
Date: 2011-11-04 15:03:03 +0000 (Fri, 04 Nov 2011)
Log Message:
-----------
Some small changes.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-04 12:51:33 UTC (rev 3372)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-04 15:03:03 UTC (rev 3373)
@@ -627,7 +627,7 @@
Set<Allocation> allocations;
for(Template t : templates){
- allocations = new HashSet<Allocation>();
+ allocations = new TreeSet<Allocation>();
for(Slot slot : t.getSlots()){
allocations = computeAllocations(slot);
@@ -639,6 +639,7 @@
slot2Allocations.put(slot, allocations);
//for tests add the property URI with http://dbpedia.org/property/ namespace
+ //TODO should be replaced by usage of a separate SOLR index
Set<Allocation> tmp = new HashSet<Allocation>();
if(slot.getSlotType() == SlotType.PROPERTY || slot.getSlotType() == SlotType.SYMPROPERTY){
for(Allocation a : allocations){
@@ -690,9 +691,7 @@
for(Allocation a : slot2Allocations.get(slot)){
for(WeightedQuery query : queries){
Query q = new Query(query.getQuery());
- if(a.getUri().equals("http://dbpedia.org/ontology/developer") && q.toString().contains("/Organisation>") && q.toString().contains("/Software>")){
- System.out.println("YES:\n" + query);
- }
+
boolean drop = false;
if(slot.getSlotType() == SlotType.PROPERTY || slot.getSlotType() == SlotType.SYMPROPERTY){
for(SPARQL_Triple triple : q.getTriplesWithVar(slot.getAnchor())){
@@ -713,17 +712,21 @@
allRanges.remove("http://www.w3.org/2002/07/owl#Thing");
String typeURI = typeTriple.getValue().getName().substring(1,typeTriple.getValue().getName().length()-1);
Set<String> allTypes = getSuperClasses(typeURI);
- allTypes.add(typeTriple.getValue().getName());
- if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer")){
- System.out.println("RANGES: " + allRanges);
- System.out.println("TYPES: " + allTypes);
- }
+ allTypes.add(typeURI);
+// if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer")){
+// System.out.println("RANGES: " + allRanges);
+// System.out.println("TYPES: " + allTypes);
+// }
if(!org.mindswap.pellet.utils.SetUtils.intersects(allRanges, allTypes)){
drop = true;
+// if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer") && q.toString().contains("/Software>")){
+// System.out.println("RANGES: " + allRanges);
+// System.out.println("TYPES: " + allTypes);
+// System.out.println("DROPPING: \n" + q.toString());
+// }
} else {
- if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer"))
- System.out.println("DROPPING: \n" + q.toString());
+
}
}
} else {
@@ -745,16 +748,16 @@
String typeURI = typeTriple.getValue().getName().substring(1,typeTriple.getValue().getName().length()-1);
Set<String> allTypes = getSuperClasses(typeURI);
allTypes.add(typeTriple.getValue().getName());
- if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer")){
- System.out.println("DOMAINS: " + allDomains);
- System.out.println("TYPES: " + allTypes);
- }
+// if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer")){
+// System.out.println("DOMAINS: " + allDomains);
+// System.out.println("TYPES: " + allTypes);
+// }
if(!org.mindswap.pellet.utils.SetUtils.intersects(allDomains, allTypes)){
drop = true;
+// System.out.println("DROPPING: \n" + q.toString());
} else {
- if(typeURI.equals("http://dbpedia.org/ontology/Organisation") && a.getUri().equals("http://dbpedia.org/ontology/developer"))
- System.out.println("DROPPING: \n" + q.toString());
+
}
}
}
@@ -816,13 +819,13 @@
*/
private Set<Allocation> computeAllocations(Slot slot){
- Set<Allocation> allocations = new HashSet<Allocation>();
+ Set<Allocation> allocations = new TreeSet<Allocation>();
SolrSearch index = getIndexBySlotType(slot);
SolrQueryResultSet rs;
for(String word : slot.getWords()){
- rs = index.getResourcesWithScores(word, 10);
+ rs = index.getResourcesWithScores(word, 30);
//debugging
// for(Iterator<SolrQueryResultItem> iter = rs.getItems().iterator();iter.hasNext();){
@@ -832,7 +835,7 @@
// }
// }
-// System.out.println(word + "->" + rs);
+ System.out.println(word + "->" + rs);
for(SolrQueryResultItem item : rs.getItems()){
int prominence = getProminenceValue(item.getUri(), slot.getSlotType());
double similarity = Similarity.getSimilarity(word, item.getLabel());
@@ -1397,7 +1400,8 @@
// String question = "Who/WP was/VBD the/DT wife/NN of/IN president/NN Lincoln/NNP";
// String question = "Who/WP produced/VBD the/DT most/JJS films/NNS";
// String question = "Which/WDT country/NN does/VBZ the/DT Airedale/NNP Terrier/NNP come/VBP from/IN";
- String question = "Which/WDT software/NN has/VBZ been/VBN developed/VBN by/IN organizations/NNS founded/VBN in/IN California/NNP";
+ String question = "When/WRB was/VBD Capcom/NNP founded/VBD";
+// String question = "Which/WDT software/NN has/VBZ been/VBN developed/VBN by/IN organizations/NNS founded/VBN in/IN California/NNP";
// String question = "How/WRB many/JJ films/NNS did/VBD Leonardo/NNP DiCaprio/NNP star/VB in/IN";
// String question = "Which/WDT music/NN albums/NNS contain/VBP the/DT song/NN Last/NNP Christmas/NNP";
SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();learner.setUseIdealTagger(true);
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-07 17:24:02
|
Revision: 3386
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3386&view=rev
Author: lorenz_b
Date: 2011-11-07 17:23:56 +0000 (Mon, 07 Nov 2011)
Log Message:
-----------
Small fix.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-07 16:07:31 UTC (rev 3385)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-07 17:23:56 UTC (rev 3386)
@@ -1402,7 +1402,10 @@
}
public int getLearnedPosition() {
- return learnedPos+1;
+ if(learnedPos >= 0){
+ return learnedPos+1;
+ }
+ return learnedPos;
}
private List<String> getResultFromLocalModel(String query, Model model){
@@ -1493,10 +1496,10 @@
// String question = "Which/WDT organizations/NNS were/VBD founded/VBN in/IN 1950/CD";
// String question = "Is/VBZ there/RB a/DT video/NN game/NN called/VBN Battle/NNP Chess/NNP";
// String question = "Which/WDT software/NN has/VBZ been/VBN developed/VBN by/IN organizations/NNS founded/VBN in/IN California/NNP";
- String question = "How/WRB many/JJ films/NNS did/VBD Leonardo/NNP DiCaprio/NNP star/VB in/IN";
+// String question = "How/WRB many/JJ films/NNS did/VBD Leonardo/NNP DiCaprio/NNP star/VB in/IN";
// String question = "Which/WDT music/NN albums/NNS contain/VBP the/DT song/NN Last/NNP Christmas/NNP";
// String question = "Which/WDT companies/NNS are/VBP located/VBN in/IN California/NNP USA/NNP";
-// String question = "Who/WP wrote/VBD the/DT book/NN The/NNP pillars/NNP of/NNP the/NNP Earth/NNP";
+ String question = "Who/WP wrote/VBD the/DT book/NN The/NNP pillars/NNP of/NNP the/NNP Earth/NNP";
SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();learner.setUseIdealTagger(true);
// SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://greententacle.techfak.uni-bielefeld.de:5171/sparql"),
// Collections.<String>singletonList(""), Collections.<String>emptyList());
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2011-11-08 13:59:20
|
Revision: 3389
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3389&view=rev
Author: lorenz_b
Date: 2011-11-08 13:59:11 +0000 (Tue, 08 Nov 2011)
Log Message:
-----------
Added cacheDir parameter to constructor.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-08 12:47:55 UTC (rev 3388)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-11-08 13:59:11 UTC (rev 3389)
@@ -24,6 +24,7 @@
import org.dllearner.algorithm.tbsl.nlp.Lemmatizer;
import org.dllearner.algorithm.tbsl.nlp.LingPipeLemmatizer;
import org.dllearner.algorithm.tbsl.nlp.PartOfSpeechTagger;
+import org.dllearner.algorithm.tbsl.nlp.StanfordPartOfSpeechTagger;
import org.dllearner.algorithm.tbsl.nlp.WordNet;
import org.dllearner.algorithm.tbsl.search.HierarchicalSolrSearch;
import org.dllearner.algorithm.tbsl.search.SolrQueryResultItem;
@@ -137,34 +138,18 @@
}
public SPARQLTemplateBasedLearner(Options options){
- init(options);
-
- Set<String> predicateFilters = new HashSet<String>();
- predicateFilters.add("http://dbpedia.org/ontology/wikiPageWikiLink");
- predicateFilters.add("http://dbpedia.org/property/wikiPageUsesTemplate");
-
- prefixMap = Prefixes.getPrefixes();
-
- modelGenenerator = new ModelGenerator(endpoint, predicateFilters);
-
- templateGenerator = new Templator();
+ this(options, new StanfordPartOfSpeechTagger());
}
public SPARQLTemplateBasedLearner(Options options, PartOfSpeechTagger tagger){
- init(options);
-
- Set<String> predicateFilters = new HashSet<String>();
- predicateFilters.add("http://dbpedia.org/ontology/wikiPageWikiLink");
- predicateFilters.add("http://dbpedia.org/property/wikiPageUsesTemplate");
-
- prefixMap = Prefixes.getPrefixes();
-
- modelGenenerator = new ModelGenerator(endpoint, predicateFilters);
-
- templateGenerator = new Templator(tagger);
+ this(options, tagger, new WordNet());
}
public SPARQLTemplateBasedLearner(Options options, PartOfSpeechTagger tagger, WordNet wordNet){
+ this(options, tagger, wordNet, "cache");
+ }
+
+ public SPARQLTemplateBasedLearner(Options options, PartOfSpeechTagger tagger, WordNet wordNet, String cacheDir){
init(options);
Set<String> predicateFilters = new HashSet<String>();
@@ -176,6 +161,7 @@
modelGenenerator = new ModelGenerator(endpoint, predicateFilters);
templateGenerator = new Templator(tagger, wordNet);
+ cache = new ExtractionDBCache(cacheDir);
}
/*
@@ -1519,8 +1505,8 @@
// String question = "Which/WDT music/NN albums/NNS contain/VBP the/DT song/NN Last/NNP Christmas/NNP";
// String question = "Which/WDT companies/NNS are/VBP located/VBN in/IN California/NNP USA/NNP";
// String question = "Who/WP wrote/VBD the/DT book/NN The/NNP pillars/NNP of/NNP the/NNP Earth/NNP";
- String question = "Who/WP is/VBZ called/VBN Dana/NNP";
- SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();learner.setUseIdealTagger(true);
+ String question = "Give me all books written by Dan Brown";
+ SPARQLTemplateBasedLearner learner = new SPARQLTemplateBasedLearner();//learner.setUseIdealTagger(true);
// SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://greententacle.techfak.uni-bielefeld.de:5171/sparql"),
// Collections.<String>singletonList(""), Collections.<String>emptyList());
SparqlEndpoint endpoint = new SparqlEndpoint(new URL("http://greententacle.techfak.uni-bielefeld.de:5171/sparql"),
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
|
From: <lor...@us...> - 2012-02-14 15:44:32
|
Revision: 3581
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3581&view=rev
Author: lorenz_b
Date: 2012-02-14 15:44:21 +0000 (Tue, 14 Feb 2012)
Log Message:
-----------
Somes updates needed because of new SOLR BOA index.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2012-02-14 10:14:06 UTC (rev 3580)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2012-02-14 15:44:21 UTC (rev 3581)
@@ -16,6 +16,7 @@
import java.util.Map.Entry;
import java.util.Set;
import java.util.SortedSet;
+import java.util.TreeMap;
import java.util.TreeSet;
import org.apache.log4j.Logger;
@@ -193,7 +194,7 @@
String boaPatternIndexUrl = options.fetch("solr.boa.properties.url");
String boaPatternIndexSearchField = options.fetch("solr.boa.properties.searchfield");
- SolrSearch patternBasedPropertyIndex = new SolrSearch(boaPatternIndexUrl, boaPatternIndexSearchField, "nlr");
+ SolrSearch patternBasedPropertyIndex = new SolrSearch(boaPatternIndexUrl, boaPatternIndexSearchField, "nlr-no-var");
//first BOA pattern then label based
// property_index = new HierarchicalSolrSearch(patternBasedPropertyIndex, labelBasedPropertyIndex);
@@ -641,18 +642,37 @@
}
private Set<WeightedQuery> getWeightedSPARQLQueries(Set<Template> templates){
+ logger.info("Generating SPARQL query candidates...");
+ Map<Slot, Set<Allocation>> slot2Allocations2 = new TreeMap<Slot, Set<Allocation>>(new Comparator<Slot>() {
+
+ @Override
+ public int compare(Slot o1, Slot o2) {
+ if(o1.getSlotType() == o2.getSlotType()){
+ return o1.getToken().compareTo(o2.getToken());
+ } else {
+ return -1;
+ }
+ }
+ });
+
+
Map<Slot, Set<Allocation>> slot2Allocations = new HashMap<Slot, Set<Allocation>>();
Set<WeightedQuery> allQueries = new TreeSet<WeightedQuery>();
Set<Allocation> allocations;
+
for(Template t : templates){
+ logger.info("Processing template:\n" + t.toString());
allocations = new TreeSet<Allocation>();
for(Slot slot : t.getSlots()){
- allocations = computeAllocations(slot, 50);
-
+ allocations = slot2Allocations2.get(slot);
+ if(allocations == null){
+ allocations = computeAllocations(slot, 20);
+ slot2Allocations2.put(slot, allocations);
+ }
slot2Allocations.put(slot, allocations);
//for tests add the property URI with http://dbpedia.org/property/ namespace
@@ -702,14 +722,14 @@
queries.addAll(tmp);
tmp.clear();
}
-
+
for(Slot slot : sortedSlots){
- if(!slot2Allocations.get(slot).isEmpty()){System.out.println(slot2Allocations.get(slot));
+ if(!slot2Allocations.get(slot).isEmpty()){
for(Allocation a : slot2Allocations.get(slot)){
for(WeightedQuery query : queries){
Query q = new Query(query.getQuery());
- boolean drop = false;
+ boolean drop = false;/*
if(slot.getSlotType() == SlotType.PROPERTY || slot.getSlotType() == SlotType.SYMPROPERTY){
for(SPARQL_Triple triple : q.getTriplesWithVar(slot.getAnchor())){
String objectVar = triple.getValue().getName();
@@ -779,7 +799,7 @@
}
}
}
- }
+ }*/
if(!drop){
@@ -811,6 +831,7 @@
}
template2Queries.put(t, qList);
}
+ logger.info("...done in ");
return allQueries;
}
@@ -859,7 +880,6 @@
// }
// }
- System.out.println(word + "->" + rs);
for(SolrQueryResultItem item : rs.getItems()){
double similarity = Similarity.getSimilarity(word, item.getLabel());
//get the labels of the redirects and compute the highest similarity
@@ -885,9 +905,11 @@
}
private Set<Allocation> computeAllocations(Slot slot, int limit){
+ logger.info("Computing allocations for " + slot);
SortedSet<Allocation> allocations = computeAllocations(slot);
if(allocations.isEmpty()){
+ logger.info("...done.");
return allocations;
}
@@ -906,7 +928,7 @@
}
}
});
-
+ logger.info("...done.");
return new TreeSet<Allocation>(l.subList(0, Math.min(limit, allocations.size())));
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X