<?xml version="1.0" encoding="UTF-8"?>
<!-- Licensed Materials - Property of IBM

     IBM SPSS Products: Text Analytics

     (C) Copyright IBM Corp. 2004, 2016

     US Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP
     Schedule Contract with IBM Corp.  -->
<spssml copyright="Copyright (c) 2006-2016 IBM Corp.,  All Rights Reserved" applicationName="TMWB" applicationVersion="1.0">
  <textService name="EXTRACTOR">
    <parameters name="DocumentSettings">
      <!-- Language: Alpha-2 ISO code -->
      <param name="Language">
        <val v="en"/>
      </param>
      <param name="MinCharPerExtract" >
        <val v="1"/>
      </param>
      <param name="MaxCharPerExtract" >
        <val v="300"/>
      </param>
      <!-- Textual Unity - 0=paragraph, 1=document -->
      <param name="TextualUnity">
        <val v="1"/>
      </param>
      <!-- Document Type: 0=Full Text, 1=Structured Text, 2=XML Text, 3=TMFC Database Text, 4=XML Database Text [0] -->
      <param name="DocumentType">
        <val v="0"/>
      </param>
      <param name="TextEncoding" >
        <val v="auto"/>
      </param>
      <param name="CheckForValidFileExtensions">
        <val v="1"/>
      </param>
    </parameters>
    <parameters name="ExtractionSettings">
      <param name="FrequencyThresHold" >
        <val v="1"/>
      </param>
      <param name="UpperCase" >
        <val v="1"/>
      </param>
      <param name="ExtractNonLingEntities">
        <val v="1"/>
      </param>
      <param name="ExtractUniterms">
        <val v="1"/>
      </param>
      <param name="ExtractPatterns">
        <val v="1"/>
      </param>
      <param name="Normalization">
        <val v="1"/>
      </param>
      <param name="MaxWordLength">
        <val v="25"/>
      </param>
      <param name="EnableCharacterCodesAsStopWords">
        <val v="0"/>
      </param>
      <param name="ListOfCharacterCodesAsStopWords">
        <val v=""/>
      </param>
      <param name="SaveMatchInRepository">
        <val v="1"/>
      </param>
      <param name="ParseSolution">
        <!--force the 1 pass solution. Extraction/Indexation will be done at same time and then grouping will be applied if other settings are compatible-->
        <!-- <val v=”1pass”> -->
        <!--force the 2 pass solution (if other settings are compatible) -->
        <!-- <val v=”2pass”> -->
        <!--auto detect if a 1 pass solution or a 2 pass solution is required regarding TLA and Fuzzy Grouping settings (default)-->
        <val v="2pass" />
      </param>
      <!-- ShowJapaneseNativeResult: to activate debug mode (show direct ntt data results) -->
      <param name="ShowJapaneseNativeResult">
        <val v="0"/>
      </param>
      <!-- DShowJapaneseNativeResultOuputFile: filename that will contains NTT Data debug results -->
      <param name="ShowJapaneseNativeResultOuputFile">
        <val v="JapaneseNativeResult.txt"/>
      </param>
      <param name="CreateMinimalModel">
        <val v="0"/>
      </param>
    </parameters>
    <parameters name="GroupingSettings">
      <param name="FuzzyGrouping">
        <val v="0"/>
      </param>
      <param name="FuzzyGroupingMinChar">
        <val v="5"/>
      </param>
      <param name="NumberOfPlainPermutedWords">
        <val v="3"/>
      </param>
      <param name="PersonGrouping">
        <val v="0"/>
      </param>
      <!-- GroupJapaneseAlgorithm: possible combinable values (case sensitive) EXCLUDE_REP_WORD, EXCLUDE_SEARCH_WORD, WORD_UNIQUE, HINSHI_CONVERSION_A -->
      <param name="GroupJapaneseAlgorithms">
        <val v="EXCLUDE_REP_WORD"/>
        <val v="EXCLUDE_SEARCH_WORD"/>
        <val v="WORD_UNIQUE"/>
        <val v="HINSHI_CONVERSION_A"/>
      </param>

    </parameters>
    <parameters name="IndexationSettings">
      <param name="FrequencyThresHold">
        <val v="1"/>
      </param>
      <!-- TLA Output Mode: default : "" (no TLA output file generated) 
         other possible value: text, debug, base, classifier, poorxml, richxml -->
      <param name="TLAOutputMode">
        <val v=""/>
      </param>
      <!-- TLA Output File: 
       File path (current path is the working directory) that will contains the TLA output (default = ptnmatcher.txt) 
       (if TLAOutputMode is not empty) -->
      <param name="TLAOutputFile">
        <val v="ptnmatcher.txt"/>
      </param>
      <!-- TLA Hit Only: For debug purpose, output sentence that have hitonly (default = 1) -->
      <param name="TLAHitOnly">
        <val v="1"/>
      </param>
      <param name="ApplyTLA">
        <val v="0"/>
      </param>
      <param name="OptimiseSpeed">
        <val v="0"/>
      </param>
    </parameters>
    <parameters name="TranslationSettings">
      <param name="LanguageWeaverHostName">
        <val v="localhost"/>
      </param>
      <param name="LanguageWeaverPort">
        <val v="9999"/>
      </param>
      <param name="TranslationAccuracy">
        <val v="2"/>
      </param>
      <param name="ReUseTranslation">
        <val v="0"/>
      </param>
      <param name="SuspendInvalidCharactersFilter">
        <val v="0"/>
      </param>
      <param name="LinkType">
        <val v="0"/>
      </param>
      <param name="LanguageWeaverSourceLanguage">
        <val v="ara"/>
      </param>
      <param name="LanguageWeaverTargetLanguage">
        <val v="eng"/>
      </param>
      <!-- TLAJapaneseAlgorithms: possible combinable values (case sensitive) : "Depend", "Sense" (taken into account only if ApplyTLA=1) -->
      <param name="TLAJapaneseAlgorithms">
        <val v="Sense"/>
      </param>
      <!-- TLAJapaneseSenseMode: possible value (case sensitive) : "SENSE_MODE_HATSUGEN", "SENSE_MODE_TANBUN" or "SENSE_MODE_ALL" -->
      <param name="TLAJapaneseSenseMode">
        <val v="SENSE_MODE_TANBUN"/>
      </param>
    </parameters>
    <parameters name="documentProcessor">
      <!-- extensio.lst-->
      <param name="fileExtensions">
        <val v=".txt"/>
        <val v=".htm"/>
        <val v=".html"/>
        <val v=".text"/>
        <val v=".shtml"/>
        <val v=".ppt"/>
        <val v=".pptx"/>
        <val v=".pptm"/>
        <val v=".doc"/>
        <val v=".docx"/>
        <val v=".docm"/>
        <val v=".pdf"/>
        <val v=".rtf"/>
        <val v=".xml"/>
        <val v=".xls"/>
        <val v=".xlsx"/>
        <val v=".xlsm"/>
        <val v=".$"/>
      </param>
      <!-- exclurep.lst-->
      <param name="excludedDirectories">
        <val v="c:\httpd\htdocs\foobarfoo\"/>
      </param>
      <!-- mtag.txt-->
      <param name="tags">
        <val v="&lt;TEXT&gt;"/>
        <val v="&lt;string&gt;"/>
      </param>
      <!-- mclef.txt-->
      <param name="textFields">
        <val v="TITLE:"/>
        <val v="ABSTRACT:"/>
      </param>
      <!-- docId.txt-->
      <param name="docIdTag">
        <val v="&lt;ID&gt;%s&lt;/ID&gt;"/>
      </param>
    </parameters>
  </textService>
</spssml>

