Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3291] trunk/components-ext/src/main/java/org/ dllearner/algorithm/tbsl
|
From: <lor...@us...> - 2011-09-29 08:47:21
|
Revision: 3291
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3291&view=rev
Author: lorenz_b
Date: 2011-09-29 08:47:15 +0000 (Thu, 29 Sep 2011)
Log Message:
-----------
Integrated new indexes in learning process.
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/SolrSearch.java
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/ThresholdSlidingSolrSearch.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 2011-09-28 18:15:59 UTC (rev 3290)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/learning/SPARQLTemplateBasedLearner.java 2011-09-29 08:47:15 UTC (rev 3291)
@@ -26,7 +26,9 @@
import org.dllearner.algorithm.qtl.util.ModelGenerator.Strategy;
import org.dllearner.algorithm.tbsl.nlp.Lemmatizer;
import org.dllearner.algorithm.tbsl.nlp.LingPipeLemmatizer;
+import org.dllearner.algorithm.tbsl.search.HierarchicalSolrSearch;
import org.dllearner.algorithm.tbsl.search.SolrSearch;
+import org.dllearner.algorithm.tbsl.search.ThresholdSlidingSolrSearch;
import org.dllearner.algorithm.tbsl.sparql.Query;
import org.dllearner.algorithm.tbsl.sparql.RatedQuery;
import org.dllearner.algorithm.tbsl.sparql.SPARQL_Prefix;
@@ -129,20 +131,22 @@
private void init(Options options){
String resourcesIndexUrl = options.fetch("solr.resources.url");
String resourcesIndexSearchField = options.fetch("solr.resources.searchfield");
- resource_index = new SolrSearch(resourcesIndexUrl, resourcesIndexSearchField);
+ resource_index = new ThresholdSlidingSolrSearch(resourcesIndexUrl, resourcesIndexSearchField, 1.0, 0.1);
String classesIndexUrl = options.fetch("solr.classes.url");
String classesIndexSearchField = options.fetch("solr.classes.searchfield");
- class_index = new SolrSearch(classesIndexUrl, classesIndexSearchField);
+ class_index = new ThresholdSlidingSolrSearch(classesIndexUrl, classesIndexSearchField, 1.0, 0.1);
String propertiesIndexUrl = options.fetch("solr.properties.url");
String propertiesIndexSearchField = options.fetch("solr.properties.searchfield");
- property_index = new SolrSearch(propertiesIndexUrl, propertiesIndexSearchField);
+ SolrSearch labelBasedPropertyIndex = new SolrSearch(propertiesIndexUrl, propertiesIndexSearchField);
String boaPatternIndexUrl = options.fetch("solr.boa.properties.url");
String boaPatternIndexSearchField = options.fetch("solr.boa.properties.searchfield");
- boa_pattern_property_index = new SolrSearch(boaPatternIndexUrl, boaPatternIndexSearchField);
+ SolrSearch patternBasedPropertyIndex = new SolrSearch(boaPatternIndexUrl, boaPatternIndexSearchField);
+ property_index = new HierarchicalSolrSearch(patternBasedPropertyIndex, labelBasedPropertyIndex);
+
int maxIndexResults = Integer.parseInt(options.fetch("solr.query.limit"), 10);
maxQueryExecutionTimeInSeconds = Integer.parseInt(options.get("sparql.query.maxExecutionTimeInSeconds", "20"));
@@ -507,41 +511,18 @@
words = pruneList(slot.getWords());
}
- if(slot.getSlotType() == SlotType.PROPERTY || slot.getSlotType() == SlotType.SYMPROPERTY){
- for(String word : words){
- tmp = new TreeSet<String>(new StringSimilarityComparator(word));
- uris = uriCache.get(word);
- index = boa_pattern_property_index;
- if(uris == null){
- uris = index.getResources(word);
- uriCache.put(word, uris);
- }
- index = property_index;
- if(uris.size() < 10){
- uris.addAll(index.getResources(word));
- }
- if(uris.size() < 10){
- uris.addAll(index.getResources("" + word + "~0.8"));
- }
- tmp.addAll(uris);
- sortedURIs.addAll(tmp);
- tmp.clear();
+ for(String word : words){
+ tmp = new TreeSet<String>(new StringSimilarityComparator(word));
+ uris = uriCache.get(word);
+
+ if(uris == null){
+ uris = index.getResources(word, 5);
+ uriCache.put(word, uris);
}
- } else {
- for(String word : words){
- tmp = new TreeSet<String>(new StringSimilarityComparator(word));
- uris = uriCache.get(word);
- if(uris == null){
- uris = index.getResources(word);
- uriCache.put(word, uris);
- }
- if(uris.size() < 10){
- uris.addAll(index.getResources("" + word + "~0.7"));
- }
- tmp.addAll(uris);
- sortedURIs.addAll(tmp);
- tmp.clear();
- }
+
+ tmp.addAll(uris);
+ sortedURIs.addAll(tmp);
+ tmp.clear();
}
slot2URI.put(slot, sortedURIs);
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/SolrSearch.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/SolrSearch.java 2011-09-28 18:15:59 UTC (rev 3290)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/SolrSearch.java 2011-09-29 08:47:15 UTC (rev 3291)
@@ -54,6 +54,10 @@
@Override
public List<String> getResources(String queryString, int limit, int offset) {
+ return findResources(queryString, limit, offset);
+ }
+
+ protected List<String> findResources(String queryString, int limit, int offset){
List<String> resources = new ArrayList<String>();
QueryResponse response;
try {
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/ThresholdSlidingSolrSearch.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/ThresholdSlidingSolrSearch.java 2011-09-28 18:15:59 UTC (rev 3290)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/search/ThresholdSlidingSolrSearch.java 2011-09-29 08:47:15 UTC (rev 3291)
@@ -28,12 +28,14 @@
double threshold = 1;
- while(resources.size() < limit){
- resources.addAll(getResources(queryString + "~" + threshold, limit - resources.size()));
+ String queryWithThreshold = queryString;
+ while(resources.size() < limit && threshold >= minThreshold){
+ if(threshold < 1){
+ queryWithThreshold = queryString + "~" + threshold;
+ }
+ System.out.println(queryWithThreshold);
+ resources.addAll(findResources(queryWithThreshold, limit - resources.size(), 0));
threshold -= step;
- if(threshold < minThreshold){
- break;
- }
}
@@ -42,4 +44,4 @@
-}
+}
\ No newline at end of file
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
