Modeling Frames in Argumentation
9217Authors
nan
Abstract
AbstractIn argumentation, framing is used to emphasize a specific aspect of a controversial topic while concealing others. When talking about legalizing drugs, for instance, its economical aspect may be emphasized. In general, we call a set of arguments that focus on the same aspect a frame. An argumentative text has to serve the “right” frame(s) to convince the audience to adopt the author’s stance (e.g., being pro or con legalizing drugs). More specifically, an author has to choose frames that fit the audience’s cultural background and interests. This paper introduces frame identification, which is the task of splitting a set of arguments into non-overlapping frames. We present a fully unsupervised approach to this task, which first removes topical information and then identifies frames using clustering. For evaluation purposes, we provide a corpus with 12, 326 debate-portal arguments, organized along the frames of the debates’ topics. On this corpus, our approach outperforms different strong baselines, achieving an F1-score of 0.28.
Keywords
Full Content
Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 2922–2932, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 2922 Modeling Frames in Argumentation Yamen Ajjour Bauhaus-Universität Weimar Faculty of Media, Webis Group yamen.ajjour@uni-weimar.de Milad Alshomary Paderborn University Department of Computer Science milad.alshomary@upb.de Henning Wachsmuth Paderborn University Department of Computer Science henningw@upb.de Benno Stein Bauhaus-Universität Weimar Faculty of Media, Webis Group benno.stein@uni-weimar.de Abstract In argumentation, framing is used to empha- size a specific aspect of a controversial topic while concealing others. When discussing the legalization of drugs, for instance, its econom- ical aspect may be emphasized. In general, we call a set of arguments that focus on the same aspect a frame. An argumentative text has to serve the “right” frame(s) to convince the au- dience to adopt the author’s stance (e.g., be- ing pro or con legalizing drugs). More specif- ically, an author has to choose frames that fit the audience’s interests and cultural back- ground. This paper introduces frame identifi- cation, which is the task of splitting a set of ar- guments into a set of non-overlapping frames. We present a fully unsupervised approach to this task, which first removes topical informa- tion from the arguments and then identifies frames using clustering. For evaluation pur- poses, we provide a corpus with 12 326 debate- portal arguments, organized along the frames of the debates’ topics. On this corpus, our ap- proach outperforms different strong baselines, achieving an F1-score of 0.28. 1 Introduction Different interests, cultural backgrounds, and so- cializations make people disagree on taking a cer- tain course of action. A debate is a means for the involved parties to resolve their disagreement. A debate is characterized by a topic, e.g., “Should one legalize abortion?”. Upon the topic, the disagreeing parties have a pro or con stance respectively, say, “Abortion should be legalized” or “Abortion should not be legalized”. A stance is supported by argu- ments; a pro argument could be “Abortion is good for a free society because it gives women their basic right of controlling their bodies.” while a con argu- ment could be “Abortion is against human rights because it is a systematic murder of innocent life.” One part of an argument (here: the part before the word “because”) is called a conclusion, the other part consists of one or more premises. A debate can be considered as set of pro and con arguments. Typically, numerous arguments exist for a topic. The parties involved in a debate have to choose among the arguments, thereby framing the topic by emphasizing a specific aspect while concealing oth- ers. We call the set of the arguments which share an aspect a frame. More specifically, a frame F is subset of a set of arguments A, F ⊆ A. Like- wise, a set of frames, {F1, . . . , Fk} covers a set of arguments iff. A ⊆ �k j Fj. For instance, …
Content truncated to 500 words
Labeled Text
AbstractIn argumentation, framing is used to emphasize a specific aspect of a controversial topic while concealing others. When talking about legalizing drugs, for instance, its economical aspect may be emphasized. In general, we call a set of arguments that focus on the same aspect a frame. An argumentative text has to serve the “right” frame(s) to convince the audience to adopt the author’s stance (e.g., being pro or con legalizing drugs). More specifically, an author has to choose frames that fit the audience’s cultural background and interests. This paper introduces frame identification, which is the task of splitting a set of arguments into non-overlapping frames. We present a fully unsupervised approach to this task, which first removes topical information and then identifies frames using clustering. For evaluation purposes, we provide a corpus with 12, 326 debate-portal arguments, organized along the frames of the debates’ topics. On this corpus, our approach outperforms different strong baselines, achieving an F1-score of 0.28.
Publication Details
- Paper ID:
- 9217
- Year:
- 2019
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan