IsaMaks / HEEM-dataset

HEEM-dataset

• guidelines (in Dutch)
• DONE human annotated data ( NAF)
• DONE machine annotated data (NAF)
• lexicon based on annotated data (LMF)
• overview annotation tags, and emotion hierarchies
• • korte beschrijving per dataset
• data license
• Translation between Dutch and English labels
• Corpus metadata?
• anything else

The dataset consists of four subsets:

1. Annotation corpus: texts selected from Nederlab manually annotaded with HEEM labels (including humor modifiers and intensifiers) (29 texts)
2. Ceneton: texts selected from Ceneton (34 texts)
3. Corpus big: other texts selected from Nederlab (149 texts)
4. EDBO: texts selected from Early Dutch Books Online (67 texts)

naf

The naf directory contains the annotations and predicted labels in NAF- format. The emotions can be found in the emotions-layer:

<emotions>
  <emotion id="emo0">
  <emotion_target/>
  <emotion_holder/>
  <emotion_expression/>
    <span>
      <!-- kop -->
      <target id="t1874"/>
    </span>
    <externalReferences>
      <externalRef reference="conceptType:bodyPart" resource="heem"/>
      <externalRef reference="head" resource="heem:bodyParts"/>
    </externalReferences>
  </emotion>
  <emotion id="emo1">
    <emotion_target/>
    <emotion_holder/>
    <emotion_expression/>
    <span>
      <!-- maek my de kop niet warm -->
      <target id="t1871"/>
      <target id="t1872"/>
      <target id="t1873"/>
      <target id="t1874"/>
      <target id="t1875"/>
      <target id="t1876"/>
    </span>
    <externalReferences>
      <externalRef reference="conceptType:bodilyProcess" resource="heem"/>
      <externalRef reference="emotionType:anger" resource="heem"/>
      <externalRef reference="anger" resource="heem:clusters"/>
      <externalRef reference="negative" resource="heem:posNeg"/>
    </externalReferences>
  </emotion>