Mining arguments from well-formed and user generated content in Social Media
2017-04-01

Teaching personnel


Lecturer

Formalia


Targeted audience
  • Komedia Master
  • Angewandte Informatik Master
  • ISE Master
  • Technomathematik Master

Description


Das Seminar beschäftigt sich mit Verfahren, die zum Argument Mining benutzt werden.

This seminar focuses on approaches used to perform argument mining.

Notes


Every participant is supposed to prepare a presentation of 40 minutes about a choosen topic from the list. A literature research for is mandatory, where relevant works (3 papers which form the state-of-the art in the field and further 3-5 papers which have similar weight in that field) will be choosen in consultation with the supervisor. The talks have to be given at the seminar. The literature is in english, but the talk can be given in german or in english. All participants are expected to take part in the discussions about the presentation topics.

After the presentation a composition should be written, which also needs to address all issues raised during the presentation days. The elaboration should not merely be a summary or translation of the underlying papers, but it should point out the advantages and disadvantages of the considered architecture or methods and classify it in the context.

Presentation-topics and supervisors


  1. Detection of argumentative texts
  2. Argument classification: claims and premises
  3. Linking claims and premises
  4. Retrieving argumentative documents
  5. Retrieving relevant premises to claims
  6. Stance detection in arguments
  7. Strategies of argumentative corpora collection
  8. Influence of corpora on the argument mining pipeline
  9. Argument mining in Twitter or other social platforms
  10. Sentiment detection in arguments
  11. Argumentative schemes
  12. Rumour detection in social media
  13. Stance detection in rumours
  14. Veracity classification for rumours
  15. Clustering of short texts
  16. Cluster labeling

Dates


The registration is now open. Send an E-Mail to Ahmet Aker including the following information: student id, first and lastname, name of subject, semester, and up to three topics of choice.

  • 3.04.2017 - 24.04.2017: Registration
  • 25.04.2017: Topic & Mentor (per mail)
  • 26.04.2017: Introduction to scientific working (14:00 - 15:30 LF 226)
  • 26.05.2017: Relevant papers
  • 09.06.2017: Thesis of the presentation
  • 23.06.2017: Structure
  • 07.07.2017: Presentation slides
  • End of semester -ish: Block seminar
  • Two weeks later: Written report

Relevant Slides


The following slides contain information that can be used as help for the seminar (sorry the slides are in German).