svn commit: r1403368 - in /stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction: engine/KeywordLinkingEngine.java impl/EntityLinker.java linking/LabelTokenizer.java

[rwesten]

Author: rwesten
Date: Mon Oct 29 15:33:25 2012
New Revision: 1403368

URL: http://svn.apache.org/viewvc?rev=1403368&view=rev
Log:
STANBOL-740: Fixed two bugs: (1) Language specific TextProcessingConfigs where 
ignored (2) FULL matches had use matched processable tokens instead of the 
covered tokens. This Caused an IllegalStateException in case of linebreaks 
within an fully matched entity.

Modified:
    
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/engine/KeywordLinkingEngine.java
    
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/impl/EntityLinker.java
    
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/linking/LabelTokenizer.java

Modified: 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/engine/KeywordLinkingEngine.java
URL: 
http://svn.apache.org/viewvc/stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/engine/KeywordLinkingEngine.java?rev=1403368&r1=1403367&r2=1403368&view=diff
==============================================================================
--- 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/engine/KeywordLinkingEngine.java
 (original)
+++ 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/engine/KeywordLinkingEngine.java
 Mon Oct 29 15:33:25 2012
@@ -455,7 +455,7 @@ public class KeywordLinkingEngine 
             log.info("    ... with language specific TextProcessingConfig");
         }
         EntityLinker entityLinker = new EntityLinker(at,language, 
-            defaultTextProcessingConfig, entitySearcher, linkerConfig, 
labelTokenizer);
+            tpc, entitySearcher, linkerConfig, labelTokenizer);
         //process
         entityLinker.process();
         //write results (requires a write lock)

Modified: 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/impl/EntityLinker.java
URL: 
http://svn.apache.org/viewvc/stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/impl/EntityLinker.java?rev=1403368&r1=1403367&r2=1403368&view=diff
==============================================================================
--- 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/impl/EntityLinker.java
 (original)
+++ 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/impl/EntityLinker.java
 Mon Oct 29 15:33:25 2012
@@ -625,6 +625,11 @@ public class EntityLinker {
                 //if(foundTokens == coveredTokens){
                 if(foundTokens == labelTokens.length && foundTokens == 
coveredTokens){
                     labelMatch = MATCH.FULL;
+                    //set found to covered: May be lower because only
+                    //processable tokens are counted, but FULL also checks
+                    //of non-processable!
+                    foundTokens = coveredTokens;
+                    foundProcessableTokens = coveredProcessableTokens;
                 } else {
                     labelMatch = MATCH.PARTIAL;
                 }

Modified: 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/linking/LabelTokenizer.java
URL: 
http://svn.apache.org/viewvc/stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/linking/LabelTokenizer.java?rev=1403368&r1=1403367&r2=1403368&view=diff
==============================================================================
--- 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/linking/LabelTokenizer.java
 (original)
+++ 
stanbol/branches/stanbol-nlp-processing/enhancer/engines/keywordextraction/src/main/java/org/apache/stanbol/enhancer/engines/keywordextraction/linking/LabelTokenizer.java
 Mon Oct 29 15:33:25 2012
@@ -11,7 +11,7 @@ public interface LabelTokenizer {
 
     /**
      * Key used to configure the languages supported for tokenizing labels.
-     * If not present the asumption is that the tokenizer supports all 
languages.
+     * If not present the assumption is that the tokenizer supports all 
languages.
      */
     String SUPPORTED_LANUAGES = 
"enhancer.engines.keywordextraction.labeltokenizer.languages";