Discrete Argument Representation Learning for Interactive Argument Pair Identification

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 5467–5478 June 6–11, 2021. ©2021 Association for Computational Linguistics 5467 Discrete Argument Representation Learning for Interactive Argument Pair Identification Lu Ji1 , Zhongyu Wei2∗, Jing Li3 , Qi Zhang1 , Xuanjing Huang1 1School of Computer Science, Fudan University, Shanghai, China 2School of Data Science, Fudan University, China 3Department of Computing, The Hong Kong Polytechnic University, Hong Kong, China {17210240034,zywei,qz,xjhuang}@fudan.edu.cn jing-amelia.li@polyu.edu.hk Abstract In this paper, we focus on identifying interac- tive argument pairs from two posts with op- posite stances to a certain topic. Consider- ing opinions are exchanged from different per- spectives of the discussing topic, we study the discrete representations for arguments to capture varying aspects in argumentation lan- guages (e.g., the debate focus and the partic- ipant behavior). Moreover, we utilize hierar- chical structure to model post-wise informa- tion incorporating contextual knowledge. Ex- perimental results on the large-scale dataset collected from CMV show that our proposed framework can significantly outperform the competitive baselines. Further analyses reveal why our model yields superior performance and prove the usefulness of our learned repre- sentations. 1 Introduction Arguments play a central role in decision mak- ing on social issues. Striving to automatically understand human arguments, computational ar- gumentation becomes a growing field in natural language processing. It can be analyzed at two levels — monological argumentation and dialog- ical argumentation. Existing research on mono- logical argumentation covers argument structure prediction (Stab and Gurevych, 2014), claims generation (Bilu and Slonim, 2016), essay scor- ing (Taghipour and Ng, 2016), etc. Recently, dia- logical argumentation becomes an active topic. In the process of dialogical arguments, partici- pants exchange arguments on a given topic (Aster- han and Schwarz, 2007; Hunter, 2013). With the popularity of online debating forums, large volume of dialogical arguments are daily formed, concern- ing wide range of topics. A social media dialogical argumentation example from ChangeMyView sub- reddit is shown in Figure 1. There we show two ∗*Corresponding author CMV: The position of vice president of the USA should be eliminated from our government. Post A: a1: .. [If..... the ........... president ..is....... either....... killed...or.......... resigns, .... the ..... vice........... president...is ..a ......... horrible......... choice ..to...... take...... over ......... office.] a2: The speaker of the House would be more qualified for the position. a3: ::: [I’m::::: willing::to::: bet::: that:::: John:::::: Boehner::::: would ::: have::: an ::::: easier :::: time :::::: dealing :::: with ::::::: congress ::as::::::: president ::: than::: Joe::::: Biden ::::: would ::: due :: to :: his::::::: constant :::::::: interaction ::: with :: it.] a4: If Boehner took office, as a republican, would he do something to veto bills Obama supported? Post B: b1:............ [Seriously,...... stop..... this.............. hyperbole.] b2:::: [Do::: you :::: think ::: that ::: have::::::: anything:: to:: do:::: with :: the::: fact::: that::::::: Boehner:is :a :::::::: republican,::: and::::::::: republicans:::::: control :::::::: congress?] b3: That argument has much less to do with the individuals than it does with the …

Content truncated to 500 words