Mining, Assessing, and Improving Arguments in NLP and the Social Sciences 

Computational argumentation is an interdisciplinary research field, connecting NLP to other disciplines such as the social sciences. This tutorial will focus on a task that recently got into the center of attention, argument quality assessment: what makes an argument good or bad? 

The tutorial will be structured along three coordinates:

1. the notions of argument quality across disciplines: how do we recognize good and bad arguments?

2. the modeling of subjectivity: who argues to whom; what are their beliefs?

3. the generation of improved arguments: what makes an argument better?

A key feature of this tutorial is its interactive nature: We will involve the participants in two annotation studies on the assessment and the improvement of quality, and we will encourage them to reflect on the challenges and potential of these tasks.

✨Scholarship Applications ✨

We are accepting applications for scholarships, consisting of a voucher to cover the registration fee for the Tutorial. The application deadline is May 13 (anywhere on earth), and we will notify recipients by May 15.

Instructors

Gabriella Lapesa, Leibniz Institute for the Social Sciences (GESIS) and Heinrich-Heine University of Düsseldorf

Eva Maria Vecchi, University of Stuttgart

Serena Villata,  Laboratoire I3S, CNRS, Sophia Antipolis, France

Henning Wachsmuth, Leibniz University Hannover

Schedule 

Morning

• 09:00 – 10:30 Session 1: Mining arguments [slides]
  
  Coffee break

• 11:00 – 12:00 Session 2: NLP Perspective: Assessing argument quality [slides]

• 12:00 – 13:00 Interactive session 1: Assessing argument quality [form]
  
  Lunch break

Afternoon

• 14:00 – 15:00 Session 4: Social sciences perspective: Assessing deliberative  [slides]

• 15:00 – 16:00 Session 5: Modeling Subjectivity [slides]
  
  Coffee break

• 16:30 – 17:00 Interactive session II: Improving argument quality [form1] [form2]

• 17:00 – 17:30 Session 7: Generation to improve argument quality [slides]

• 17:30 – 18:00 Session 8: Wrapping Up [slides]
  

Recommended readings

• Computational Argumentation Quality Assessment in Natural Language (Wachsmuth et al. 2017)

• Five Years of Argument Mining: a Data-driven Analysis (Cabrio and Villata, 2018)

• Towards Argument Mining for Social Good: a Survey (Vecchi et al. 2021)

• Argument Quality Assessment in the Age of Instruction-Following Large Language Models (Wachsmuth et al. 2024)