Identifying Argument Components through TextRank

Abstract

In this paper we examine the application of an unsupervised extractive summarisation algorithm, TextRank, on a different task, the identification of argumentative components. Our main motivation is to examine whether there is any potential overlap between extractive summarisation and argument mining, and whether approaches used in summarisation (which typically model a document as a whole) can have a positive effect on tasks of argument mining. Evaluation has been performed on two corpora containing user posts from an on-line debating forum and persuasive essays. Evaluation results suggest that graph-based approaches and approaches targeting extractive summarisation can have a positive effect on tasks related to argument mining.