Exploring Discourse Structures for Argument Impact Classification

TL;DR

This paper investigates how discourse structures influence argument impact classification and introduces DisCOC, a model that integrates discourse relations with contextual features to improve persuasive power prediction.

Contribution

It is the first to explicitly incorporate discourse relations in argument impact classification and demonstrates the effectiveness of structural discourse information with large-scale language models.

Findings

Discourse relations are crucial for identifying argument impact.

DisCOC improves classification performance over baseline models.

Attention and gate mechanisms effectively model context and discourse structures.

Abstract

Discourse relations among arguments reveal logical structures of a debate conversation. However, no prior work has explicitly studied how the sequence of discourse relations influence a claim's impact. This paper empirically shows that the discourse relations between two arguments along the context path are essential factors for identifying the persuasive power of an argument. We further propose DisCOC to inject and fuse the sentence-level structural discourse information with contextualized features derived from large-scale language models. Experimental results and extensive analysis show that the attention and gate mechanisms that explicitly model contexts and texts can indeed help the argument impact classification task defined by Durmus et al. (2019), and discourse structures among the context path of the claim to be classified can further boost the performance.