LPAttack: A Feasible Annotation Scheme for Capturing Logic Pattern of Attacks in Arguments

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2446–2459 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2446 LPAttack: A Feasible Annotation Scheme for Capturing Logic Pattern of Attacks in Arguments Farjana Sultana Mim1, Naoya Inoue2,3,∗, Shoichi Naito1,3,4, Keshav Singh1, Kentaro Inui1,3 1Tohoku University, 2Stony Brook University 3RIKEN, 4Ricoh Company, Ltd. naoya.inoue.lab@gmail.com, inui@tohoku.ac.jp {mim.farjana.sultana.t3, naito.shoichi.t1, singh.keshav.t4}@dc.tohoku.ac.jp Abstract In argumentative discourse, persuasion is often achieved by refuting or attacking others’ arguments. Attacking an argument is not always straightforward and often consists of complex rhetorical moves in which arguers may agree with a logic of an argument while attacking another logic. Furthermore, an arguer may neither deny nor agree with any logics of an argument, instead ignore them and attack the main stance of the argument by providing new logics and presupposing that the new logics have more value or importance than the logics presented in the attacked argument. However, there are no studies in computational argumentation that capture such complex rhetorical moves in attacks or the presuppositions or value judgments in them. To address this gap, we introduce LPAttack, a novel annotation scheme that captures the common modes and complex rhetorical moves in attacks along with the implicit presuppositions and value judgments. Our annotation study shows moderate inter-annotator agreement, indicating that human annotation for the proposed scheme is feasible. We publicly release our annotated corpus and the annotation guidelines. Keywords: argument, counterargument, attack, logic, argumentation, reasoning, annotation, pattern, debate 1. Introduction Argumentation plays a central role in human commu- nication, where refuting or attacking others’ arguments is a common persuasion strategy (Walton et al., 2010). Attacks in arguments can have different modes (Wal- ton, 2009; Cramer and Guillaume, 2018) and often con- sist of complex rhetorical moves, e.g., one might attack the conclusion (i.e., the statement expressing position or belief of the arguer) while agreeing with a premise (i.e., statement providing support or reason for the con- clusion) of the argument (Afantenos and Asher, 2014). Consider an example debate in Fig. 1, where oppos- ing teams give two argumentative speeches. In the debate, the counterargument (CA) does not deny the premise of the initial argument (IA), i.e., death penalty deprives the chance of rehabilitation of the criminals. Instead, she implicitly agrees with it while denying the conclusion of the IA, i.e., death penalty should be abolished by giving more importance or value to the death penalty than the rehabilitation of the crimi- nals. Although this value judgment is implicit in the CA speech, CA explicitly provides a reason behind her value judgment (bold text in CA). Automatically identifying such internal logic patterns can help a wide range of natural language processing (NLP) applica- tions. For example, in an educational domain, this can help machines diagnose learners’ arguments and pro- vide feedback to the learners. Prior studies in NLP that focused on attacks in argu- ments mainly worked on the classification of argumen- tative relations (e.g., support, attack, neutral), identify- *Present affiliation: …

Content truncated to 500 words