Classification and Clustering of Arguments with Contextualized Word Embeddings

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 567–578 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 567 Classification and Clustering of Arguments with Contextualized Word Embeddings Nils Reimers, Benjamin Schiller, Tilman Beck, Johannes Daxenberger, Christian Stab, Iryna Gurevych Ubiquitous Knowledge Processing Lab (UKP-TUDA) Department of Computer Science, Technische Universit¨at Darmstadt www.ukp.tu-darmstadt.de Abstract We experiment with two recent contextual- ized word embedding methods (ELMo and BERT) in the context of open-domain argu- ment search. For the first time, we show how to leverage the power of contextual- ized word embeddings to classify and cluster topic-dependent arguments, achieving impres- sive results on both tasks and across multiple datasets. For argument classification, we im- prove the state-of-the-art for the UKP Senten- tial Argument Mining Corpus by 20.8 percent- age points and for the IBM Debater - Evidence Sentences dataset by 7.4 percentage points. For the understudied task of argument clus- tering, we propose a pre-training step which improves by 7.8 percentage points over strong baselines on a novel dataset, and by 12.3 per- centage points for the Argument Facet Simi- larity (AFS) Corpus.1 1 Introduction Argument mining methods have been applied to different tasks such as identifying reasoning struc- tures (Stab and Gurevych, 2014), assessing the quality of arguments (Wachsmuth et al., 2017), or linking arguments from different documents (Cabrio and Villata, 2012). Broadly speaking, ex- isting methods either approach argument mining from the discourse-level perspective (aiming to an- alyze local argumentation structures), or from an information-seeking perspective (aiming to detect arguments relevant to a predefined topic). While discourse-level approaches mostly focus on the analysis of single documents or document col- lections (Eger et al., 2017), information-seeking approaches need to be capable of dealing with heterogeneous sources and topics (Shnarch et al., 2018) and also face the problem of redundancy, as 1Code and models available: https://github.com/UKPLab/ acl2019-BERT-argument-classification-and-clustering arguments might be repeated across sources. As a result, this perspective naturally calls for a subse- quent clustering step, which is able to identify and aggregate similar arguments for the same topic. In this work, we focus on the latter perspective, re- ferring to it as open-domain argument search, and show how contextualized word embeddings can be leveraged to overcome some of the challenges in- volved in topic-dependent argument classification and clustering. Identifying arguments for unseen topics is a challenging task for machine learning systems. The lexical appearance for two topics, e.g. “net neutrality” and “school uniforms”, is vastly dif- ferent. Hence, in order to perform well, systems must develop a deep semantic understanding of both the topic as well as the sources to search for arguments. Even more so, clustering similar argu- ments is a demanding task, as fine-grained seman- tic nuances may determine whether two arguments (talking about the same topic) are similar. Figure 1 gives an example of arguments on the topic “net neutrality”. Both arguments center around the as- pect of “equal access for every Internet user” but are differently phrased. …

Content truncated to 500 words