Generating Informative Conclusions for Argumentative Texts

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3482–3493 August 1–6, 2021. ©2021 Association for Computational Linguistics 3482 Generating Informative Conclusions for Argumentative Texts Shahbaz Syed † Khalid Al-Khatib † Milad Alshomary ‡ Henning Wachsmuth ‡ Martin Potthast † †Leipzig University ‡Paderborn University <shahbaz.syed@uni-leipzig.de> Abstract The purpose of an argumentative text is to sup- port a certain conclusion. Yet, they are often omitted, expecting readers to infer them rather. While appropriate when reading an individual text, this rhetorical device limits accessibility when browsing many texts (e.g., on a search engine or on social media). In these scenarios, an explicit conclusion makes for a good candi- date summary of an argumentative text. This is especially true if the conclusion is informative, emphasizing specific concepts from the text. With this paper we introduce the task of gen- erating informative conclusions: First, Webis- ConcluGen-21 is compiled, a large-scale cor- pus of 136,996 samples of argumentative texts and their conclusions. Second, two paradigms for conclusion generation are investigated; one extractive, the other abstractive in nature. The latter exploits argumentative knowledge that augment the data via control codes and finetun- ing the BART model on several subsets of the corpus. Third, insights are provided into the suitability of our corpus for the task, the differ- ences between the two generation paradigms, the trade-off between informativeness and con- ciseness, and the impact of encoding argumen- tative knowledge. The corpus, code, and the trained models are publicly available.1 1 Introduction A conclusion of an argument is a statement that con- veys a stance towards a specific target (Bar-Haim et al., 2017; Alshomary et al., 2020b). Drawing conclusions is an integral part of argumentation, but often various conclusions may be drawn from a set of premises. Consider the following argumen- tative text on caffeine adapted from the web:2 “Caffeine stimulates the nervous system, sig- naling fat cells to break down body fat. It also 1https://github.com/webis-de/ACL-21 2https://www.healthline.com/nutrition/top-13-evidence- based-health-benefits-of-coffee increases epinephrine (adrenaline) levels, a fight- or-flight hormone preparing the body for physical exertion. With free body fat acids as fuel, on aver- age, 12% higher performance is attainable.” Consider further these alternative conclusions: 1. Caffeine is good. 2. Caffeine improves physical performance. The first conclusion conveys a pro stance towards the target, caffeine. The second, conveys a pro stance towards caffeine, too, but it also emphasizes a specific concept (“physical performance”). The former conclusion is generic, only indicating the stance, while the latter is informative; a distinction also made in text summarization (Section 3).3 Argumentative texts include short arguments, such as forum posts and reviews, as well as long- form texts, such as essays, blogs, and editorials. Most of these typically have an intended conclu- sion of which the authors seek to persuade their readers.4 While the conclusion may be already implied in a given text, authors often choose not to explicitly provide one, either for rhetorical rea- sons (Habernal and Gurevych, 2015; Al-Khatib et al., 2016), or to encourage critical thinking (Mar- tin et al., 2003). However, …

Content truncated to 500 words