Stance Classification using Dialogic Properties of Persuasion
23855Authors
nan
Abstract
Public debate functions as a forum for both expressing and forming opinions, an impor- tant aspect of public life. We present results for automatically classifying posts in online debate as to the position, or STANCE that the speaker takes on an issue, such as Pro or Con. We show that representing the dialogic struc- ture of the debates in terms of agreement rela- tions between speakers, greatly improves per- formance for stance classification, over mod- els that operate on post content and parent- post context alone. 1
Keywords
Full Content
2012 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 592–596, Montr´eal, Canada, June 3-8, 2012. c⃝2012 Association for Computational Linguistics Stance Classification using Dialogic Properties of Persuasion Marilyn A. Walker, Pranav Anand, Robert Abbott and Ricky Grant Baskin School of Engineering & Linguistics Department University of California Santa Cruz Santa Cruz, Ca. 95064, USA maw,panand,abbott,rgrant@soe.ucsc.edu Abstract Public debate functions as a forum for both expressing and forming opinions, an impor- tant aspect of public life. We present results for automatically classifying posts in online debate as to the position, or STANCE that the speaker takes on an issue, such as Pro or Con. We show that representing the dialogic struc- ture of the debates in terms of agreement rela- tions between speakers, greatly improves per- formance for stance classification, over mod- els that operate on post content and parent- post context alone. 1 Introduction Public debate functions as a forum for both express- ing and forming opinions. Three factors affect opin- ion formation, e.g. the perlocutionary uptake of de- bate arguments (Cialdini, 2000; Petty and Cacioppo, 1988; Petty et al., 1981). First, there is the ARGU- MENT itself, i.e. the propositions discussed along with the logical relations between them. Second is the SOURCE of the argument (Chaiken, 1980), e.g. the speaker’s expertise, or agreement relations be- tween speakers. The third factor consists of proper- ties of the AUDIENCE such as prior beliefs, social identity, personality, and cognitive style (Davies, 1998). Perlocutionary uptake in debates primar- ily occurs in the audience, who may be undecided, while debaters typically express a particular position or STANCE on an issue, e.g. Pro or Con, as in the online debate dialogues in Figs. 1, 2, and 3. Previous computational work on debate covers three different debate settings: (1) congressional de- Post Stance Utterance P1 PRO I feel badly for your ignorance because although there maybe a sliver of doubt that mankind may have evolved from previous animals, there is no doubt that the Earth and the cosmos have gone through evolution and are continuing to do so P2 CON As long as there are people who doubt evolu- tion, both lay and acedamia, then evolution is in doubt. And please don’t feel bad for me. I am perfectly secure in my “ignorance”. P3 PRO By that measure, as long as organic chemistry, physics and gravity are in doubt by both lay and acedamia, then organic chemistry, physics and gravity are in doubt. Gravity is a theory. Why aren’t you giving it the same treatment you do to evolution? Or is it because you are ignorant? Angelic Falling anyone? P4 CON I’m obviously ignorant. Look how many times i’ve been given the title. “Gravity is a theory. Why aren’t you giving it the same treatment you do to evolution?” Because it doesn’t carry the same weight. ;P Figure 1: All posts linked via rebuttal links. The topic was “Evolution”, with sides ”Yes, I Believe” vs. “No, I Dont …
Content truncated to 500 words
Labeled Text
Public debate functions as a forum for both expressing and forming opinions, an important aspect of public life. We present results for automatically classifying posts in online debate as to the position, or STANCE that the speaker takes on an issue, such as Pro or Con. We show that representing the dialogic structure of the debates in terms of agreement relations between speakers, greatly improves performance for stance classification, over models that operate on post content and parentpost context alone. 1
Publication Details
- Paper ID:
- 23855
- Year:
- 2012
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan