To Revise or Not to Revise: Learning to Detect Improvable Claims for Argumentative Writing Support

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 15799–15816 July 9-14, 2023 ©2023 Association for Computational Linguistics To Revise or Not to Revise: Learning to Detect Improvable Claims for Argumentative Writing Support Gabriella Skitalinskaya1,2 and Henning Wachsmuth1 1 Institute of Artificial Intelligence, Leibniz University Hannover 2 Department of Computer Science, University of Bremen {g.skitalinska,h.wachsmuth}@ai.uni-hannover.de Abstract Optimizing the phrasing of argumentative text is crucial in higher education and professional development. However, assessing whether and how the different claims in a text should be revised is a hard task, especially for novice writers. In this work, we explore the main chal- lenges to identifying argumentative claims in need of specific revisions. By learning from col- laborative editing behaviors in online debates, we seek to capture implicit revision patterns in order to develop approaches aimed at guiding writers in how to further improve their argu- ments. We systematically compare the ability of common word embedding models to capture the differences between different versions of the same text, and we analyze their impact on various types of writing issues. To deal with the noisy nature of revision-based corpora, we pro- pose a new sampling strategy based on revision distance. Opposed to approaches from prior work, such sampling can be done without em- ploying additional annotations and judgments. Moreover, we provide evidence that using con- textual information and domain knowledge can further improve prediction results. How use- ful a certain type of context is, depends on the issue the claim is suffering from, though. 1 Introduction Text revision is an essential part of professional writing and is typically a recursive process until a somehow optimal phrasing is achieved from the author’s point of view. Aside from proofreading and copyediting, text revision subsumes substan- tive and rhetorical changes not only at the lexical, syntactic, and semantic levels, but also some that may require knowledge about the topic of discus- sion or about conventions of the domain or genre. An optimal phrasing is especially important in ar- gumentative writing, where it is considered a key component in academic and professional success: An argument’s style directly affects its persuasive effect on the audience (El Baff et al., 2020). Cell phones cause brain cancer. Bitcoins makes tax evasion quite simple. Open-Source licenses are a mess. Cell phone radiation causes brain cancer. Bitcoins make tax evasion quite simple. Licensing options for open-source software are complex. Cell phone radiation may cause cancer. Cryptocurrencies enable tax evasion. Specification Grammar Reframing Reframing Clarification Is claim optimal? If not, how to improve? v1 v2 v3 Figure 1: Examples of revision histories of argumen- tative claims from the online debate platform Kialo. Colors denote claims considered optimal (light green) and claims requiring revisions (medium red). We seek to identify if and how a new claim should be revised. But how to know whether an argument is phrased well enough and no more revisions are needed? Most existing approaches to argument quality as- sessment score arguments on …

Content truncated to 500 words