|
Abstract:
|
It is often interesting to know how similar two persons argue, e.g. when comparing the attitudes of voters and political parties, or when building an argumentation-based recommender system. Those applications need a distance function, which should give intuitive results. In this paper, we present seven functions which calculate how similar the attitudes of two agents are in an argumentation. We evaluate how good those functions match the results of a human baseline which we determined in a previous work. As it turns out, variants of the p-metric, Cosine, and Soergel distance best agree with human intuition.
|
