svn commit: r1425333 - in /stanbol/trunk/enhancement-engines: pom.xml smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties

[rw...@apache.org]

Author: rwesten
Date: Sat Dec 22 18:42:06 2012
New Revision: 1425333

URL: http://svn.apache.org/viewvc?rev=1425333&view=rev
Log:
STANBOL-860: added metadata.properties; added smartcn engine to reactor pom

Modified:
    stanbol/trunk/enhancement-engines/pom.xml
    stanbol/trunk/enhancement-engines/smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties

Modified: stanbol/trunk/enhancement-engines/pom.xml
URL: http://svn.apache.org/viewvc/stanbol/trunk/enhancement-engines/pom.xml?rev=1425333&r1=1425332&r2=1425333&view=diff
==============================================================================
--- stanbol/trunk/enhancement-engines/pom.xml (original)
+++ stanbol/trunk/enhancement-engines/pom.xml Sat Dec 22 18:42:06 2012
@@ -63,6 +63,7 @@
     <module>opennlp-pos</module>
     <module>opennlp-ner</module>
     <module>opennlp-chunker</module>
+    <module>smartcn-token</module> <!-- Chinese sentence detection & tokenizer -->
     <module>nlp2rdf</module> <!-- converts AnalyzedText ContentPart to RDF -->
   
   	<!-- EntityLinking -->

Modified: stanbol/trunk/enhancement-engines/smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties
URL: http://svn.apache.org/viewvc/stanbol/trunk/enhancement-engines/smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties?rev=1425333&r1=1425332&r2=1425333&view=diff
==============================================================================
--- stanbol/trunk/enhancement-engines/smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties (original)
+++ stanbol/trunk/enhancement-engines/smartcn-token/src/main/resources/OSGI-INF/metatype/metatype.properties Sat Dec 22 18:42:06 2012
@@ -16,11 +16,11 @@
 #  under the License.
 #
 
-org.apache.stanbol.enhancer.engines.opennlp.token.impl.OpenNlpTokenizerEngine.name=Apache \
-Stanbol Enhancer Engine: OpenNLP Tokenizer
+org.apache.stanbol.enhancer.engines.smartcn.impl.SmartcnTokenizerEngine.name=Apache \
+Stanbol Enhancer Engine: Smartcn Tokenizer
 org.apache.stanbol.enhancer.engines.opennlp.token.impl.OpenNlpTokenizerEngine.description=Enhancement \
-Engine that tokenizes text by using the OpenNLP Tokenizer. It can be configured to use a \
-custom TokenizerModel or the SimpleTokenizer by configuring 'SIMPLE' as modle name.
+Engine that detect sentences and tokenizes Chinese text by using the Solr/Lucene \
+smartcn analyzers.
 
 
 stanbol.enhancer.engine.name.name=Name
@@ -29,13 +29,3 @@ used in the RESTful interface '/engine/<
 service.ranking.name=Ranking
 service.ranking.description=If two enhancement engines with the same name are active the \
 one with the higher ranking will be used to process parsed content items.
-
-org.apache.stanbol.enhancer.token.languages.name=Language configuration
-org.apache.stanbol.enhancer.token.languages.description=Takes a list of ISO \
-  language codes. '*' is the Wildcard; '!{lang}' to exclude a language; \
-  '{lang};model={tokenizer-model-file-name}' to configure a \
-  custom OpenNLP tokenizer model for a language. The name 'SIMPLE' can be used to \
-  force the usage of the SimpleTokenizer for a lnaugage. \
-  Models are loaded via the Stanbol DataFileProvider service. So users can e.g. \
-  put models in the datafiles directory \
-  (defaults to '{stanbol-working-dir}/stanbol/datafiles')