folia_sonar.xml 9.51 KB
<?xml version="1.0" encoding="UTF-8" ?>
<mtas>

  <!-- START MTAS INDEX CONFIGURATION -->
  <index>
    <!-- START GENERAL SETTINGS MTAS INDEX PROCESS -->
    <payload index="false" />
    <offset index="false" />
    <realoffset index="false" />
    <parent index="true" />
    <!-- END GENERAL SETTINGS MTAS INDEX PROCESS -->
  </index>
  <!-- END MTAS INDEX CONFIGURATION -->


    <!-- START CONFIGURATION MTAS FOLIA PARSER -->
    <parser name="mtas.analysis.parser.MtasFoliaParser">

      <!-- START GENERAL SETTINGS MTAS FOLIA PARSER -->
      <autorepair value="true" />
      <makeunique value="true" />
      <!-- END GENERAL SETTINGS MTAS FOLIA PARSER -->

      <!-- START REFERENCES -->
      <references>
        <reference name="wref" ref="id" />
      </references>
      <!-- END REFERENCES -->

      <!-- START MAPPINGS -->
      <mappings>

        <!-- START WORDS -->
        <mapping type="word" name="w">
        </mapping>
        <mapping type="word" name="w">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
          <condition>
            <item type="attribute" name="class" />
            <item type="attribute" name="class" not="true" condition="WORD" />
          </condition>
        </mapping>
        <!-- END WORDS -->

        <!-- START WORD ANNOTATIONS -->
        <mapping type="wordAnnotation" name="t">
          <token type="string" offset="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="text" />
            </post>
          </token>
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
              <item type="string" value="_lc" />
            </pre>
            <post>
              <item type="text" filter="ascii,lowercase" />
            </post>
          </token>
          <condition>
            <item type="ancestor" number="0" />
            <item type="ancestorWord" number="1" />
            <item type="unknownAncestor" number="0" />
          </condition>
        </mapping>
        <mapping type="wordAnnotation" name="lemma">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
          <condition>
            <item type="attribute" name="class" />
            <item type="ancestor" number="0" />
            <item type="unknownAncestor" number="0" />
            <item type="attribute" name="set" condition="http://ilk.uvt.nl/folia/sets/frog-mblem-nl" />
          </condition>
        </mapping>
        <mapping type="wordAnnotation" name="morphology">
        </mapping>
        <mapping type="wordAnnotation" name="morpheme">
          <condition>
            <item type="ancestor" number="1" />
            <item type="ancestorName" condition="morphology" />
          </condition>
        </mapping>
        <mapping type="wordAnnotation" name="t">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="ancestorName" />
            </pre>
            <post>
              <item type="text" />
            </post>
          </token>
          <condition>
            <item type="ancestorName" distance="0" condition="morpheme" />
          </condition>
        </mapping>
        <mapping type="wordAnnotation" name="pos">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="head" />
            </post>
            <payload>
              <item type="attribute" name="confidence" />
            </payload>
          </token>
          <condition>
            <item type="ancestor" number="0" />
            <item type="unknownAncestor" number="0" />
            <item type="attribute" name="class" />
            <item type="attribute" name="set" condition="http://ilk.uvt.nl/folia/sets/frog-mbpos-cgn" />          
          </condition>
        </mapping>
        <mapping type="wordAnnotation" name="feat">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
              <item type="attribute" name="subset" prefix="." />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
            <payload>
              <item type="ancestorAttribute" distance="0" name="confidence" />
            </payload>
          </token>
          <condition>
            <item type="ancestor" number="1" />
            <item type="unknownAncestor" number="0" />
            <item type="attribute" name="class" />
            <item type="attribute" name="subset" />
            <item type="ancestorAttribute" name="set" condition="http://ilk.uvt.nl/folia/sets/frog-mbpos-cgn" />
          </condition>
        </mapping>
        <!-- END WORD ANNOTATIONS -->

        <!-- START RELATIONS -->
        <mapping type="relation" name="chunk">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
            <payload>
              <item type="attribute" name="confidence" />
            </payload>
          </token>
          <condition>
            <item type="attribute" name="class" />
          </condition>
        </mapping>
        <mapping type="relation" name="dependency">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
          <condition>
            <item type="attribute" name="class" />
          </condition>
        </mapping>
        <mapping type="relation" name="hd">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="ancestorName" distance="0" />
              <item type="name" prefix="." />
            </pre>
          </token>
          <condition>
            <item type="ancestorName" condition="dependency" />
            <item type="ancestor" number="1" />
          </condition>
        </mapping>
        <mapping type="relation" name="dep">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="ancestorName" distance="0" />
              <item type="name" prefix="." />
            </pre>
          </token>
          <condition>
            <item type="ancestor" number="1" />
            <item type="ancestorName" condition="dependency" />
          </condition>
        </mapping>
        <mapping type="relation" name="entities">
        </mapping>
        <mapping type="relation" name="entity">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
            <payload>
              <item type="attribute" name="confidence" />
            </payload>
          </token>
          <condition>
            <item type="ancestor" number="1" />
            <item type="ancestorName" condition="entities" />
          </condition>
        </mapping>
        <!-- END RELATIONS -->

        <!-- START GROUPS -->
        <mapping type="group" name="s">
          <token type="string" offset="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
        </mapping>
        <mapping type="group" name="p">
          <token type="string" offset="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
        </mapping>
        <mapping type="group" name="div">
          <token type="string" offset="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
        </mapping>
        <mapping type="group" name="head">
          <token type="string" offset="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
        </mapping>
        <!-- END GROUPS -->

        <!-- START GROUP ANNOTATIONS -->
        <mapping type="groupAnnotation" name="lang">
          <token type="string" offset="false" realoffset="false" parent="false">
            <pre>
              <item type="name" />
            </pre>
            <post>
              <item type="attribute" name="class" />
            </post>
          </token>
        </mapping>
        <!-- END GROUP ANNOTATIONS -->

      </mappings>
      <!-- END MAPPINGS -->

    </parser>
    <!-- END CONFIGURATION MTAS FOLIA PARSER -->


</mtas>