Merge with 8358b3eb6e5b5237ade42438836b869e6141a776

HeidelTime · Apr 2, 2013 · ceaba2e · ceaba2e
2 parents b57e9ee + 3cd8946
commit ceaba2e
Show file tree

Hide file tree

Showing 75 changed files with 3,072 additions and 11 deletions.
diff --git a/desc/annotator/StanfordPOSTaggerWrapper.xml b/desc/annotator/StanfordPOSTaggerWrapper.xml
@@ -0,0 +1,95 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<analysisEngineDescription xmlns="http://uima.apache.org/resourceSpecifier">
+  <frameworkImplementation>org.apache.uima.java</frameworkImplementation>
+  <primitive>true</primitive>
+  <annotatorImplementationName>de.unihd.dbs.uima.annotator.stanfordtagger.StanfordPOSTaggerWrapper</annotatorImplementationName>
+  <analysisEngineMetaData>
+    <name>StanfordPOSTaggerWrapper</name>
+    <description/>
+    <version>1.0</version>
+    <vendor>UniHD DBS</vendor>
+    <configurationParameters searchStrategy="language_fallback">
+      <configurationParameter>
+        <name>model_path</name>
+        <description>Path to a model for use with the Stanford POS Tagger</description>
+        <type>String</type>
+        <multiValued>false</multiValued>
+        <mandatory>true</mandatory>
+      </configurationParameter>
+      <configurationParameter>
+        <name>config_path</name>
+        <description>Optional path to a configuration for use with the Stanford POS Tagger</description>
+        <type>String</type>
+        <multiValued>false</multiValued>
+        <mandatory>false</mandatory>
+      </configurationParameter>
+      <configurationParameter>
+        <name>annotate_tokens</name>
+        <description>Whether or not to annotate tokens</description>
+        <type>Boolean</type>
+        <multiValued>false</multiValued>
+        <mandatory>false</mandatory>
+      </configurationParameter>
+      <configurationParameter>
+        <name>annotate_sentences</name>
+        <description>Whether or not to annotate sentences.</description>
+        <type>Boolean</type>
+        <multiValued>false</multiValued>
+        <mandatory>false</mandatory>
+      </configurationParameter>
+      <configurationParameter>
+        <name>annotate_partofspeech</name>
+        <description>Whether or not to annotate part of speech (POS) information</description>
+        <type>Boolean</type>
+        <multiValued>false</multiValued>
+        <mandatory>false</mandatory>
+      </configurationParameter>
+    </configurationParameters>
+    <configurationParameterSettings>
+      <nameValuePair>
+        <name>annotate_tokens</name>
+        <value>
+          <boolean>true</boolean>
+        </value>
+      </nameValuePair>
+      <nameValuePair>
+        <name>annotate_sentences</name>
+        <value>
+          <boolean>true</boolean>
+        </value>
+      </nameValuePair>
+      <nameValuePair>
+        <name>annotate_partofspeech</name>
+        <value>
+          <boolean>true</boolean>
+        </value>
+      </nameValuePair>
+      <nameValuePair>
+        <name>model_path</name>
+        <value>
+          <string>/opt/stanford-postagger-full/models/english-bidirectional-distsim.tagger</string>
+        </value>
+      </nameValuePair>
+    </configurationParameterSettings>
+    <typeSystemDescription>
+      <imports>
+        <import location="../type/HeidelTime_TypeSystem.xml"/>
+      </imports>
+    </typeSystemDescription>
+    <typePriorities/>
+    <fsIndexCollection/>
+    <capabilities>
+      <capability>
+        <inputs/>
+        <outputs/>
+        <languagesSupported/>
+      </capability>
+    </capabilities>
+    <operationalProperties>
+      <modifiesCas>true</modifiesCas>
+      <multipleDeploymentAllowed>true</multipleDeploymentAllowed>
+      <outputsNewCASes>false</outputsNewCASes>
+    </operationalProperties>
+  </analysisEngineMetaData>
+  <resourceManagerConfiguration/>
+</analysisEngineDescription>
diff --git a/metadata/setenv b/metadata/setenv
@@ -6,5 +6,7 @@ export CLASSPATH=$DKPRO_HOME/bin:$CLASSPATH
 export CLASSPATH=$DKPRO_HOME/lib:$CLASSPATH
 ### setting the environment for jvntextpro (required for the wrapper)
 export CLASSPATH=$JVNTEXTPRO_HOME:$CLASSPATH
+### setting the environment for the stanford pos tagger (required for the wrapper)
+export CLASSPATH=$STANFORDTAGGER:$CLASSPATH
 
 
diff --git a/resources/arabic/normalization/resources_normalization_normApprox4Dates.txt b/resources/arabic/normalization/resources_normalization_normApprox4Dates.txt
@@ -0,0 +1,20 @@
+// author: Jannik Strötgen
+// email:  [email protected]
+// date:   2011-09-12
+// This file contains "approximate words" and their normalized expressions
+// for dates/times according to TIMEX3 format.
+// For example, the normalized value of "about" is "APPROX"
+// FORMAT: "approximate-word","normalized-approximate-word"
+// about
+"حوالي","APPROX"
+"نحو","APPROX"
+"تقريبا[ً]?","APPROX"
+"حول","APPROX"S
+"حتى","EQUAL_OR_LESS"
+"على ال[اأ]قل","EQUAL_OR_MORE"
+"[اأ]كثر من","MORE_THAN"
+"[اأ]قل من","LESS_THAN"
+"[اأ]طول من","MORE_THAN"
+"[اأ]قصر من","LESS_THAN"
+"في غضون","APPROX"
+"قرابة","APPROX"
diff --git a/resources/arabic/normalization/resources_normalization_normApprox4Durations.txt b/resources/arabic/normalization/resources_normalization_normApprox4Durations.txt
@@ -0,0 +1,20 @@
+// author: Jannik Strötgen
+// email:  [email protected]
+// date:   2011-09-12
+// This file contains "approximate words" and their normalized expressions
+// for dates/times according to TIMEX3 format.
+// For example, the normalized value of "about" is "APPROX"
+// FORMAT: "approximate-word","normalized-approximate-word"
+"حوالي","APPROX"
+"نحو","APPROX"
+"تقريبا[ً]?","APPROX"
+"حول","APPROX"
+"حتى","EQUAL_OR_LESS"
+"على ال[اأ]قل","EQUAL_OR_MORE"
+"[اأ]كثر من","MORE_THAN"
+"[اأ]قل من","LESS_THAN"
+"[اأ]طول من","MORE_THAN"
+"[اأ]قصر من","LESS_THAN"
+"في غضون","APPROX"
+"قرابة","APPROX"
+
diff --git a/resources/arabic/normalization/resources_normalization_normArabicDigit.txt b/resources/arabic/normalization/resources_normalization_normArabicDigit.txt
@@ -0,0 +1,17 @@
+// author: Jannik Strötgen
+// email: [email protected]
+// date:  2011-06-10
+// This file contains "duration numbers" and their normalized expressions
+// according to TIMEX3 format.
+// For example, the normalized value of "one" is "1"
+// FORMAT: "duration-number","normalized-duration-number"
+"\u0660","0"
+"\u0661","1"
+"\u0662","2"
+"\u0663","3"
+"\u0664","4"
+"\u0665","5"
+"\u0666","6"
+"\u0667","7"
+"\u0668","8"
+"\u0669","9"
diff --git a/resources/arabic/normalization/resources_normalization_normDateNumber.txt b/resources/arabic/normalization/resources_normalization_normDateNumber.txt
@@ -0,0 +1,123 @@
+// author: Jannik Strötgen
+// email: [email protected]
+// date:  2011-06-10
+// This file contains "duration numbers" and their normalized expressions
+// according to TIMEX3 format.
+// For example, the normalized value of "one" is "1"
+// FORMAT: "duration-number","normalized-duration-number"
+"واحد[هة]?","01"
+"[اإ]ثن[ت]?[اي]ن","02"
+"ثلاث[هة]?","03"
+"[اأ]ربع[هة]?","04"
+"خمس[هة]?","05"
+"ست[هة]?","06"
+"سبع[هة]?","07"
+"ثماني[هة]?","08"
+"تسع[هة]?","09"
+"عشر[هة]?","10"
+"[أاإ]حد[ى]?[\s]?عشر[هة]?","11"
+"[اإ]ثن[ت]?[اي][\s]?عشر[هة]?","12"
+"ثلاث[هة]?[\s]?عشر[هة]?","13"
+"[أا]ربع[هة]?[\s]?عشر[هة]?","14"
+"خمس[هة]?[\s]?عشر[هة]?","15"
+"ست[هة]?[\s]?عشر[هة]?","16"
+"سبع[هة]?[\s]?عشر[هة]?","17"
+"ثماني[هة]?[\s]?عشر[هة]?","18"
+"تسع[هة]?[\s]?عشر[هة]?","19"
+"عشر[وي]ن","20"
+"[اإ]حدى[\s]?و[\s]?عشر[وي]ن","21"
+"واحد[\s]?و[\s]?عشر[وي]ن","21"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?عشر[وي]ن","22"
+"ثلاث[هة]?[\s]?و[\s]?عشر[وي]ن","23"
+"[أا]ربع[هة]?[\s]?و[\s]?عشر[وي]ن","24"
+"خمس[هة]?[\s]?و[\s]?عشر[وي]ن","25"
+"ست[هة]?[\s]?و[\s]?عشر[وي]ن","26"
+"سبع[هة]?[\s]?و[\s]?عشر[وي]ن","27"
+"ثماني[هة]?[\s]?و[\s]?عشر[وي]ن","28"
+"تسع[هة]?[\s]?و[\s]?عشر[وي]ن","29"
+"ثلاث[وي]ن","30"
+"[اإ]حدى[\s]?و[\s]?ثلاث[وي]ن","31"
+"واحد[\s]?و[\s]?ثلاث[وي]ن","31"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?ثلاث[وي]ن","32"
+"ثلاث[هة]?[\s]?و[\s]?ثلاث[وي]ن","33"
+"[أا]ربع[هة]?[\s]?و[\s]?ثلاث[وي]ن","34"
+"خمس[هة]?[\s]?و[\s]?ثلاث[وي]ن","35"
+"ست[هة]?[\s]?و[\s]?ثلاث[وي]ن","36"
+"سبع[هة]?[\s]?و[\s]?ثلاث[وي]ن","37"
+"ثماني[هة]?[\s]?و[\s]?ثلاث[وي]ن","38"
+"تسع[هة]?[\s]?و[\s]?ثلاث[وي]ن","39"
+"[اأ]ربع[وي]ن","40"
+"[اإ]حدى[\s]?و[\s]?[اأ]ربع[وي]ن","41"
+"واحد[\s]?و[\s]?[اأ]ربع[وي]ن","41"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?[اأ]ربع[وي]ن","42"
+"ثلاث[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","43"
+"[أا]ربع[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","44"
+"خمس[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","45"
+"ست[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","46"
+"سبع[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","47"
+"ثماني[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","48"
+"تسع[هة]?[\s]?و[\s]?[اأ]ربع[وي]ن","49"
+"خمس[وي]ن","50"
+"[اإ]حدى[\s]?و[\s]?خمس[وي]ن","51"
+"واحد[\s]?و[\s]?خمس[وي]ن","51"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?خمس[وي]ن","52"
+"ثلاث[هة]?[\s]?و[\s]?خمس[وي]ن","53"
+"[أا]ربع[هة]?[\s]?و[\s]?خمس[وي]ن","54"
+"خمس[هة]?[\s]?و[\s]?خمس[وي]ن","55"
+"ست[هة]?[\s]?و[\s]?خمس[وي]ن","56"
+"سبع[هة]?[\s]?و[\s]?خمس[وي]ن","57"
+"ثماني[هة]?[\s]?و[\s]?خمس[وي]ن","58"
+"تسع[هة]?[\s]?و[\s]?خمس[وي]ن","59"
+"ست[وي]ن","60"
+"[اإ]حدى[\s]?و[\s]?ست[وي]ن","61"
+"واحد[\s]?و[\s]?ست[وي]ن","61"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?ست[وي]ن","62"
+"ثلاث[هة]?[\s]?و[\s]?ست[وي]ن","63"
+"[أا]ربع[هة]?[\s]?و[\s]?ست[وي]ن","64"
+"خمس[هة]?[\s]?و[\s]?ست[وي]ن","65"
+"ست[هة]?[\s]?و[\s]?ست[وي]ن","66"
+"سبع[هة]?[\s]?و[\s]?ست[وي]ن","67"
+"ثماني[هة]?[\s]?و[\s]?ست[وي]ن","68"
+"تسع[هة]?[\s]?و[\s]?ست[وي]ن","69"
+"سبع[وي]ن","70"
+"[اإ]حدى[\s]?و[\s]?سبع[وي]ن","71"
+"واحد[\s]?و[\s]?سبع[وي]ن","71"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?سبع[وي]ن","72"
+"ثلاث[هة]?[\s]?و[\s]?سبع[وي]ن","73"
+"[أا]ربع[هة]?[\s]?و[\s]?سبع[وي]ن","74"
+"خمس[هة]?[\s]?و[\s]?سبع[وي]ن","75"
+"ست[هة]?[\s]?و[\s]?سبع[وي]ن","76"
+"سبع[هة]?[\s]?و[\s]?سبع[وي]ن","77"
+"ثماني[هة]?[\s]?و[\s]?سبع[وي]ن","78"
+"تسع[هة]?[\s]?و[\s]?سبع[وي]ن","79"
+"ثمان[وي]ن","80"
+"[اإ]حدى[\s]?و[\s]?ثمان[وي]ن","81"
+"واحد[\s]?و[\s]?ثمان[وي]ن","81"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?ثمان[وي]ن","82"
+"ثلاث[هة]?[\s]?و[\s]?ثمان[وي]ن","83"
+"[أا]ربع[هة]?[\s]?و[\s]?ثمان[وي]ن","84"
+"خمس[هة]?[\s]?و[\s]?ثمان[وي]ن","85"
+"ست[هة]?[\s]?و[\s]?ثمان[وي]ن","86"
+"سبع[هة]?[\s]?و[\s]?ثمان[وي]ن","87"
+"ثماني[هة]?[\s]?و[\s]?ثمان[وي]ن","88"
+"تسع[هة]?[\s]?و[\s]?ثمان[وي]ن","89"
+"تسع[وي]ن","90"
+"[اإ]حدى[\s]?و[\s]?تسع[وي]ن","91"
+"واحد[\s]?و[\s]?تسع[وي]ن","91"
+"[اإ]ثن[ت]?[اي][ن]?[\s]?و[\s]?تسع[وي]ن","92"
+"ثلاث[هة]?[\s]?و[\s]?تسع[وي]ن","93"
+"[أا]ربع[هة]?[\s]?و[\s]?تسع[وي]ن","94"
+"خمس[هة]?[\s]?و[\s]?تسع[وي]ن","95"
+"ست[هة]?[\s]?و[\s]?تسع[وي]ن","96"
+"سبع[هة]?[\s]?و[\s]?تسع[وي]ن","97"
+"ثماني[هة]?[\s]?و[\s]?تسع[وي]ن","98"
+"تسع[هة]?[\s]?و[\s]?تسع[وي]ن","99"
+"م[ا]?ئ[هة]","1"
+"م[ا]?ئت[اي][ن]?","2"
+"ثلاث[هة]?[\s]?م[ا]?ئ[هة]","3"
+"[اأ]ربع[هة]?[\s]?م[ا]?ئ[هة],"4"
+"خمس[هة]?[\s]?م[ا]?ئ[هة]","5"
+"ست[هة]?[\s]?م[ا]?ئ[هة]","6"
+"سبع[هة]?[\s]?م[ا]?ئ[هة]","7"
+"ثماني[هة]?[\s]?م[ا]?ئ[هة]","8"
+"تسع[هة]?[\s]?م[ا]?ئ[هة]","9"
diff --git a/resources/arabic/normalization/resources_normalization_normDateWord.txt b/resources/arabic/normalization/resources_normalization_normDateWord.txt
@@ -0,0 +1,32 @@
+// author: Jannik Strötgen
+// email:  [email protected]
+// date:   2011-06-10
+// This file contains "date words" and their normalized expressions
+// according to TIMEX3 format.
+// For example, the normalized value of "tomorrow" is "UNDEF-next-day"
+// FORMAT: "date-word","normalized-date-word"
+"[اأ]ول [اأ]مس","UNDEF-this-day-MINUS-2"
+"[اأ]لبارح[هة]","UNDEF-last-day"
+"(ب)?[اأ](لأ)?(لا)?مس","UNDEF-last-day"
+"[اأ]ليوم","UNDEF-this-day"
+"[اأ][ل]?غد","UNDEF-next-day"
+"غدا","UNDEF-next-day"
+"[اأ]ل[اآ]ن","PRESENT_REF"
+"في اللحظ[هة]","PRESENT_REF"
+"في [اأ]قرب وقت","FUTURE_REF"
+"قريبا[ً]?","FUTURE_REF"
+"حديثا[ً]?","PRESENT_REF"
+"حاليا[ً]?","PRESENT_REF"
+"سابقا[ً]?","PAST_REF"
+"في الماضي","PAST_REF"
+"في السابق","PAST_REF"
+"مؤخرا[ً]?","PAST_REF"
+"آنذاك","PAST_REF"
+"في الراهن","PRESENT_REF"
+"الوقت الراهن","PRESENT_REF"
+"الوقت الماضي","PAST_REF"
+"الوقت السابق","PAST_REF"
+"الوقت الحالي","PRESENT_REF"
+"هذه ال[اأ]يام","PRESENT_REF"
+"ال[اأ]يام ال[أا]خير[هة]","PAST_REF"
+