Publikationen und Abschlussarbeiten am Lehrstuhl für Rechnernetze
Do I Argue Like Them? A Human Baseline for Comparing Attitudes in Argumentations
Authors
Markus Brenneis
Martin Mauve
Title
Do I Argue Like Them? A Human Baseline for Comparing Attitudes in Argumentations
Published
In Proceedings,
Published in
Proceedings of the Workshop on Advances In Argumentation In Artificial Intelligence 2020
Keywords
argumentation graphs, metric, human baseline
Abstract
When different persons exchange their attitudes in an argumentation, it is useful to measure how similar the attitudes in that argumentation are: for distance-based recommender systems which suggest other arguments, finding clusters of similar people, or comparing one's own attitude with the attitudes of political parties. Those applications need a mathematical metric to calculate the distance between two attitudes in an argumentation. But which properties should an intuitive metric fulfill? We surveyed untrained persons to find out which properties such a metric should have to produce results matching human intuition. For that, we formulated several hypotheses for useful properties, which we then tried to confirm in our survey. As a result, we were able to identify some properties a metric for comparing attitudes represented as argumentation graphs should have, and some properties where further research is needed.
