A Dataset of General-Purpose Rebuttal

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 5591–5601, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 5591 A Dataset of General-Purpose Rebuttal Matan Orbach1, Yonatan Bilu1, Ariel Gera1, Yoav Kantor1, Lena Dankin1, Tamar Lavee1, Lili Kotlerman∗ 2, Shachar Mirkin∗3, Michal Jacovi1, Ranit Aharonov1 and Noam Slonim1 1IBM Research, 2Intuition Robotics, 3Digimind {matano,yonatanb,arielge,yoavka,lenad,tamarl, michal.jacovi, ranita,noams}@il.ibm.com, lili@intuitionrobotics.com, shachar.mirkin@digimind.com Abstract In Natural Language Understanding, the task of response generation is usually focused on responses to short texts, such as tweets or a turn in a dialog. Here we present a novel task of producing a critical response to a long argu- mentative text, and suggest a method based on general rebuttal arguments to address it. We do this in the context of the recently-suggested task of listening comprehension over argu- mentative content: given a speech on some specified topic, and a list of relevant argu- ments, the goal is to determine which of the arguments appear in the speech. The gen- eral rebuttals we describe here (written in En- glish) overcome the need for topic-specific ar- guments to be provided, by proving to be ap- plicable for a large set of topics. This allows creating responses beyond the scope of topics for which specific arguments are available. All data collected during this work is freely avail- able for research1. 1 Introduction A key element in argumentation is rebuttal: the ability to contest an argument by presenting a counter-argument. It is an important skill, not easily learned, and valued in many fields such as politics and science. It is useful for advancing your own views and beliefs over opposing ones, but perhaps more importantly, it facilitates a criti- cal examination of the views and beliefs that you hold. An automatic rebuttal system could there- fore be useful whenever critical analysis of written or spoken content is required – be it an elementary school student writing an essay or a seasoned jour- nalist composing an op-ed. In the context of Natural Language Understand- ing, the study of rebuttal and counter-arguments ∗This work was done at IBM Research. 1https://www.research.ibm.com/haifa/ dept/vst/debating_data.shtml has focused on elucidating such relations between given arguments. Indeed, such “attack” relations are the foundation of Argumentation Frameworks (Dung, 1995); such frameworks have been one of the main objects of study in computational argu- mentation. A related task, that of generating a response which need not be a rebuttal or even argumenta- tive, has been the subject of much research, espe- cially in the context of dialog systems, chat bots, and question answering. In this line of work the response typically follows a short input text, often only a sentence or two. Here we suggest the task of producing a rebuttal in response to a long argumentative text. Specif- ically, we consider spoken speeches around four minutes long. In addition to being longer, and per- haps because they are so, these kinds of texts tend to …

Content truncated to 500 words