Skip to content

cl-tohoku/CALSA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 

Repository files navigation

Designing Logic Pattern Templates for Counter-Argument Logical Structure Analysis

Currently only the formatted dataset is uploaded, other additional material will be added soon.

Contents

  • train.json: training set containing 516 CAs with only single-template label per sentence.
  • dev.json: dev set containing 128 CAs with only single-template label per sentence.
  • test.json: test set containing 134 CAs with at least one multi-template label sentence, the other sentences have single-template labels.

The meaning of each field in the json files

  • pm_id: id for the initial-argument.
  • lo_id: id for the counter-argument.
  • pm_speech: essay of an initial-argument.
  • lo_speech: essay of a counter-argument, divided by sentences.
  • ptns: template label for each sentence.
  • segments: segments for different templates. A segment is considered as one or more consecutive sentences with the same template.
    • key for each segment: template id.
    • range: start_index, end_index of of counter-argument sentences in lo_speech.
    • slots: slot-fillers for the corresponding template.

In the test set, we do not have a segment field since there are multiple ways of deciding segments due to the presence of multi-template labels. Therefore, the test set has the following fields instead.

  • slots: non-expert annotators raw annotations of slot-fillers for the corresponding CA.
  • workers_raw_annotations: non-expert annotators raw annotations of logic templates for corresponding CA.
  • filtered_slots: slot-fillers filtered from the raw slots based on the rules 1) discarding slots that are cannot be exactly found from the CA essay; 2) discarding slots that subset of other slots.

Template id - name mapping

  • 1: Mitigation (Mig)
  • 2: Alternative (Alt)
  • 3: No evidence (No Evi)
  • 4: Another true cause (ATC)
  • 5: Missing mechanism #1 (MM1)
  • 6: Missing mechanism #2 (MM2)
  • 7: No need to address (NNA)
  • 8: Negative effect due to y (Neg eff)
  • 9: Positive effects of a different perspective from y #1 (Dif Per1)
  • 10: Positive effects of a different perspective from y #2 (Dif Per2)
  • 11: Others
  • 100: Non-argumentative

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published