Explaining Controversy on Social Media via Stance Summarization

Abstract

In an era in which new controversies rapidly emerge and evolve on social media, navigating social media platforms to learn about a new controversy can be an overwhelming task. In this light, there has been significant work that studies how to identify and measure controversy online. However, we currently lack a tool for effectively understanding controversy in social media. For example, users have to manually examine postings to find the arguments of conflicting stances that make up the controversy.In this paper, we study methods to generate a stance-aware summary that explains a given controversy by collecting arguments of two conflicting stances. We focus on Twitter and treat stance summarization as a ranking problem of finding the top k tweets that best summarize the two conflicting stances of a controversial topic. We formalize the characteristics of a good stance summary and propose a ranking model accordingly. We first evaluate our methods on five controversial topics on Twitter. Our user evaluation shows that our methods consistently outperform other baseline techniques in generating a summary that explains the given controversy.

Target-adaptive Graph for Cross-target Stance Detection

“…favor, against, or none etc.) expressed in text towards a specific target [1,9,23,25,39]. Thanks in part to the availability of data sufficiently annotated with target-dependent stance labels, previous methods achieved promising performance in targetdependent stance detection when trained and tested on the same dataset of targets [7,18].…”

Analysis and Prediction of Multilingual Controversy on Reddit

“…In practice, this allows moderators and mediators to intervene timely and resolve conflicts or to advise users to include references backing up their claims in debates on social media. Hence, the prediction of controversy has been studied extensively in existing research [7,17,19,23,24,31,37,38]. However, most of these studies focused on English content and platforms such as Twitter [15,16,19,26] or Wikipedia [7,8,28,38].…”

“…Hence, the prediction of controversy has been studied extensively in existing research [7,17,19,23,24,31,37,38]. However, most of these studies focused on English content and platforms such as Twitter [15,16,19,26] or Wikipedia [7,8,28,38]. Research question.…”

“…In this work, we study controversy on the multilingual social news aggregation website Reddit, which our community has not yet widely analyzed with respect to controversy. In particular, we ask how commonly studied discussion features, such as word usage [17,23,31], writing style [17,19,31], sentiment [7,23,38], as well as user involvement [24,37] are predictive of controversy on Reddit and whether their predictive properties carry over to languages other than English. Approach.…”

“…We base our analysis on well-established features of social media users and postings previously studied in settings different to ours [7,17,19,23,24,31,37,38]. In particular, we study word usage in comments and perform subgroup discovery to find words that have been distinctively used in controversial and non-controversial comments.…”