A News Editorial Corpus for Mining Argumentation Strategies

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 3433–3443, Osaka, Japan, December 11-17 2016. A News Editorial Corpus for Mining Argumentation Strategies Khalid Al-Khatib, Henning Wachsmuth, Johannes Kiesel, Matthias Hagen, Benno Stein Faculty of Media, Bauhaus-Universität Weimar, Germany <firstname>.<lastname>@uni-weimar.de Abstract Many argumentative texts, and news editorials in particular, follow a specific strategy to persuade their readers of some opinion or attitude. This includes decisions such as when to tell an anecdote or where to support an assumption with statistics, which is reflected by the composition of differ- ent types of argumentative discourse units in a text. While several argument mining corpora have recently been published, they do not allow the study of argumentation strategies due to incom- plete or coarse-grained unit annotations. This paper presents a novel corpus with 300 editorials from three diverse news portals that provides the basis for mining argumentation strategies. Each unit in all editorials has been assigned one of six types by three annotators with a high Fleiss’ κ agreement of 0.56. We investigate various challenges of the annotation process and we conduct a first corpus analysis. Our results reveal different strategies across the news portals, exemplifying the benefit of studying editorials—a so far underresourced text genre in argument mining. 1 Introduction News editorials define a genre of written argumentative discourse whose main goal is persuasiveness. In a news editorial, an author states and defends a thesis that conveys his or her stance on a controversial topic usually related to the public interest. Editorials do not only aim to persuade readers of some opinion, but they often also propagate particular ideologies or recommend certain attitudes to the community, e.g., a specific action towards an upcoming event (van Dijk, 1992). To achieve persuasion, a news editorial follows a particular argumentation strategy that the author expects to be most suitable for the target audience, i.e., the author composes a series of claims, assumptions, and different types of evidence while using argumentative language and structure (van Dijk, 1995). This does not only cover the resort to quantitative features of text (e.g., related to lexical style, cohesion, or rhetorical structure), but it also refers to the roles, positions, flows, and relations of specific argumentative discourse units. The study of news editorials is beneficial for several tasks, such as the creation of persuasive writing strategies for writing assistance systems and qualitative media content analysis. At the same time, the rapid expansion of online news portals increases the need for algorithms that can analyze an editorial’s discourse automatically. The needed analyses include argumentation mining and evidence detection, both of which are studied in computational argumentation, an emergent area of computational linguistics. While several text corpora for such analyses have recently been published for different domains and genres, a respective resource with news editorials is missing to this day. Moreover, existing corpora stick to coarse-grained and/or incomplete annotations of the units of an argumentative discourse (see Section 2 for details), which renders the mining of an …

Content truncated to 500 words