Revision: 16334
http://gate.svn.sourceforge.net/gate/?rev=16334&view=rev
Author: adamfunk
Date: 2012-11-27 11:13:41 +0000 (Tue, 27 Nov 2012)
Log Message:
-----------
No language code -> blank rather than "___" now.
Removing some more project-specific cruft.
Modified Paths:
--------------
gate/trunk/plugins/TermRaider/applications/termraider-eng.gapp
gate/trunk/plugins/TermRaider/applications/termraider-multilingual.gapp
gate/trunk/plugins/TermRaider/resources/groovy/ApplyLanguage.groovy
gate/trunk/plugins/TermRaider/src/gate/termraider/bank/AbstractTermbank.java
gate/trunk/plugins/TermRaider/src/gate/termraider/util/AbstractBank.java
gate/trunk/plugins/TermRaider/src/gate/termraider/util/Term.java
Modified: gate/trunk/plugins/TermRaider/applications/termraider-eng.gapp
===================================================================
--- gate/trunk/plugins/TermRaider/applications/termraider-eng.gapp
2012-11-27 10:54:07 UTC (rev 16333)
+++ gate/trunk/plugins/TermRaider/applications/termraider-eng.gapp
2012-11-27 11:13:41 UTC (rev 16334)
@@ -2,7 +2,7 @@
<urlList class="gate.util.persistence.CollectionPersistence">
<localList>
<gate.util.persistence.PersistenceManager-URLHolder>
- <urlString>$relpath$../</urlString>
+ <urlString>$relpath$../../TermRaider/</urlString>
</gate.util.persistence.PersistenceManager-URLHolder>
<gate.util.persistence.PersistenceManager-URLHolder>
<urlString>$relpath$../../ANNIE/</urlString>
@@ -19,25 +19,19 @@
<application class="gate.groovy.ScriptableControllerPersistence">
<controlScript class="string">import gate.termraider.*;
-String instNamespace =
gate.termraider.util.AbstractBank.generateInstanceNamespace();
-
// equivalent to old preprocessor
eachDocument {
docReset();
tokenizer();
- //stopwords();
splitter();
- //textCat();
annieGazetteer();
posTagger();
lemmatizer();
annieJape();
orthomatcher();
- //stopwordsJape();
selectTokens();
multiwordJape();
deduplicateMW();
- applyLanguage();
}
// create the TfIdfTermbank LR
@@ -45,7 +39,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace
].toFeatureMap())
eachDocument {
@@ -58,7 +51,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace,
inputScoreFeature:'localAugTfIdf'
].toFeatureMap())
@@ -67,7 +59,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace,
inputHeadFeatures:'head;canonical',
inputAnnotationFeature:'canonical'
].toFeatureMap())
@@ -691,62 +682,6 @@
<localMap/>
</features>
</gate.util.persistence.PRPersistence>
- <gate.util.persistence.PRPersistence>
- <runtimeParams class="gate.util.persistence.MapPersistence">
- <mapType>gate.util.SimpleFeatureMapImpl</mapType>
- <localMap>
- <entry>
- <string>outputASName</string>
- <null/>
- </entry>
- <entry>
- <string>document</string>
- <null/>
- </entry>
- <entry>
- <string>corpus</string>
- <null/>
- </entry>
- <entry>
- <string>scriptParams</string>
- <gate.util.persistence.MapPersistence>
- <mapType>gate.util.SimpleFeatureMapImpl</mapType>
- <localMap>
- <entry>
- <string>defaultLanguage</string>
- <string>english</string>
- </entry>
- </localMap>
- </gate.util.persistence.MapPersistence>
- </entry>
- <entry>
- <string>inputASName</string>
- <null/>
- </entry>
- </localMap>
- </runtimeParams>
- <resourceType>gate.groovy.ScriptPR</resourceType>
- <resourceName>applyLanguage</resourceName>
- <initParams class="gate.util.persistence.MapPersistence">
- <mapType>gate.util.SimpleFeatureMapImpl</mapType>
- <localMap>
- <entry>
- <string>encoding</string>
- <string>UTF-8</string>
- </entry>
- <entry>
- <string>scriptURL</string>
- <gate.util.persistence.PersistenceManager-URLHolder>
-
<urlString>$relpath$../resources/groovy/ApplyLanguage.groovy</urlString>
- </gate.util.persistence.PersistenceManager-URLHolder>
- </entry>
- </localMap>
- </initParams>
- <features class="gate.util.persistence.MapPersistence">
- <mapType>gate.util.SimpleFeatureMapImpl</mapType>
- <localMap/>
- </features>
- </gate.util.persistence.PRPersistence>
<gate.util.persistence.LanguageAnalyserPersistence>
<runtimeParams class="gate.util.persistence.MapPersistence">
<mapType>gate.util.SimpleFeatureMapImpl</mapType>
@@ -931,4 +866,4 @@
</localMap>
</features>
</application>
-</gate.util.persistence.GateApplication>
+</gate.util.persistence.GateApplication>
\ No newline at end of file
Modified:
gate/trunk/plugins/TermRaider/applications/termraider-multilingual.gapp
===================================================================
--- gate/trunk/plugins/TermRaider/applications/termraider-multilingual.gapp
2012-11-27 10:54:07 UTC (rev 16333)
+++ gate/trunk/plugins/TermRaider/applications/termraider-multilingual.gapp
2012-11-27 11:13:41 UTC (rev 16334)
@@ -22,8 +22,6 @@
<application class="gate.groovy.ScriptableControllerPersistence">
<controlScript class="string">import gate.termraider.*;
-String instNamespace =
gate.termraider.util.AbstractBank.generateInstanceNamespace();
-
// equivalent to old preprocessor
eachDocument {
docReset();
@@ -48,7 +46,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace
].toFeatureMap())
eachDocument {
@@ -61,7 +58,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace,
inputScoreFeature:'localAugTfIdf'
].toFeatureMap())
@@ -70,7 +66,6 @@
corpora:[corpus],
inputAnnotationTypes:'SingleWord;MultiWord',
debugMode:true,
- namespaceBase:instNamespace,
inputHeadFeatures:'head;canonical',
inputAnnotationFeature:'canonical'
].toFeatureMap())
Modified: gate/trunk/plugins/TermRaider/resources/groovy/ApplyLanguage.groovy
===================================================================
--- gate/trunk/plugins/TermRaider/resources/groovy/ApplyLanguage.groovy
2012-11-27 10:54:07 UTC (rev 16333)
+++ gate/trunk/plugins/TermRaider/resources/groovy/ApplyLanguage.groovy
2012-11-27 11:13:41 UTC (rev 16334)
@@ -22,9 +22,6 @@
String language = sentence.getFeatures().get("lang").toString();
candidate.getFeatures().put("lang", language);
}
- else if (defaultLanguage != null) {
- candidate.getFeatures().put("lang", defaultLanguage);
- }
}
}
}
Modified:
gate/trunk/plugins/TermRaider/src/gate/termraider/bank/AbstractTermbank.java
===================================================================
---
gate/trunk/plugins/TermRaider/src/gate/termraider/bank/AbstractTermbank.java
2012-11-27 10:54:07 UTC (rev 16333)
+++
gate/trunk/plugins/TermRaider/src/gate/termraider/bank/AbstractTermbank.java
2012-11-27 11:13:41 UTC (rev 16334)
@@ -35,7 +35,6 @@
protected String inputASName;
protected Set<String> inputAnnotationTypes;
protected String docFreqProperty;
- protected String namespaceBase;
// transient to allow serialization
protected transient List<Action> actionsList;
@@ -298,23 +297,9 @@
}
- /* The following parameters are just used for ontology and CSV generation */
- /* Set this in the application for consistency between termbanks;
- * otherwise it will be randomly generated. */
- @CreoleParameter(comment = "Namespace base (including '#') for ontology
generation",
- defaultValue = "")
- public void setNamespaceBase(String nsb) {
- this.namespaceBase = nsb;
- }
-
- public String getNamespaceBase() {
- return this.namespaceBase;
- }
-
-
/* Default value is overridden in the implementations */
- @CreoleParameter(comment = "name of ontology doc frequency property",
+ @CreoleParameter(comment = "name of doc frequency property",
defaultValue = "")
public void setDocFreqProperty(String name) {
this.docFreqProperty = name;
Modified:
gate/trunk/plugins/TermRaider/src/gate/termraider/util/AbstractBank.java
===================================================================
--- gate/trunk/plugins/TermRaider/src/gate/termraider/util/AbstractBank.java
2012-11-27 10:54:07 UTC (rev 16333)
+++ gate/trunk/plugins/TermRaider/src/gate/termraider/util/AbstractBank.java
2012-11-27 11:13:41 UTC (rev 16334)
@@ -29,12 +29,8 @@
public abstract class AbstractBank extends AbstractLanguageResource {
private static final long serialVersionUID = -9168657973312733783L;
- public static final String randomNsBase =
"http://www.gate.ac.uk/ns/arcomem/";
-
-
protected Set<String> languages, types;
-
public abstract Double getMinScore();
public abstract Double getMaxScore();
@@ -122,11 +118,5 @@
public Set<Corpus> getCorpora() {
return this.corpora;
}
-
- public static String generateInstanceNamespace() {
- return Utilities.generateID(randomNsBase, "#");
- }
-
-
}
Modified: gate/trunk/plugins/TermRaider/src/gate/termraider/util/Term.java
===================================================================
--- gate/trunk/plugins/TermRaider/src/gate/termraider/util/Term.java
2012-11-27 10:54:07 UTC (rev 16333)
+++ gate/trunk/plugins/TermRaider/src/gate/termraider/util/Term.java
2012-11-27 11:13:41 UTC (rev 16334)
@@ -21,10 +21,8 @@
public class Term implements Comparable<Term>, Serializable {
private static final long serialVersionUID = -4849144989013687570L;
- public static final String LANGUAGE_ERROR_CODE = "___";
private String termString, languageCode, type;
- private boolean typed;
private int hashCode;
private String toString;
@@ -36,14 +34,7 @@
this.setup();
}
- public Term(String termString, String languageCode) {
- this.termString = termString;
- this.languageCode = languageCode;
- this.type = "";
- this.setup();
- }
-
public Term(Annotation annotation, Document document, String
languageFeature,
String stringFeature) {
this.type = annotation.getType();
@@ -54,25 +45,21 @@
private void setup() {
- if (type == null) {
- type = "";
+ if (languageCode == null) {
+ languageCode = "";
}
- typed = ! type.isEmpty();
- hashCode = termString.hashCode() + languageCode.hashCode();
- if (typed) {
- hashCode = termString.hashCode() + languageCode.hashCode() +
type.hashCode();
- toString = termString + " (" + languageCode + "," + type + ")";
+
+ hashCode = termString.hashCode() + languageCode.hashCode() +
type.hashCode();
+
+ if (languageCode.isEmpty()) {
+ toString = termString + " (" + type + ")";
}
else {
- hashCode = termString.hashCode() + languageCode.hashCode();
- toString = termString + " (" + languageCode + ")";
+ toString = termString + " (" + languageCode + "," + type + ")";
}
}
- public boolean typed() {
- return typed;
- }
-
+
public String toString() {
return toString;
}
@@ -119,7 +106,7 @@
public static String getLanguage(Annotation annotation, String
languageFeature) {
- String language = LANGUAGE_ERROR_CODE;
+ String language = "";
if (annotation.getFeatures().containsKey(languageFeature)) {
language = annotation.getFeatures().get(languageFeature).toString();
}
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov
_______________________________________________
GATE-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gate-cvs