A Lightweight Approach for User and Keyword Classification in Controversial Topics
Ahmad Zareie, Kalina Bontcheva, Carolina Scarton
TL;DR
This work tackles the challenge of classifying user stances on controversial topics in OSNs without relying on labelled data or post content. It introduces the Lightweight Random Walk Method (LRM), which builds a hashtag-sharing graph and applies a tailored local random walk to infer hashtag stances from a single seed per class, subsequently deriving user classifications from hashtag intensities. Across UK General Election 2019 and Brexit Twitter datasets, LRM outperforms several baselines in macro F1 while offering competitive runtime and the ability to track stance evolution over time. The approach is language- and platform-agnostic, data-efficient, and directly useful for social scientists and policymakers seeking scalable, real-time insight into public opinion dynamics.
Abstract
Classifying the stance of individuals on controversial topics and uncovering their concerns is crucial for social scientists and policymakers. Data from Online Social Networks (OSNs), which serve as a proxy to a representative sample of society, offers an opportunity to classify these stances, discover society's concerns regarding controversial topics, and track the evolution of these concerns over time. Consequently, stance classification in OSNs has garnered significant attention from researchers. However, most existing methods for this task often rely on labelled data and utilise the text of users' posts or the interactions between users, necessitating large volumes of data, considerable processing time, and access to information that is not readily available (e.g. users' followers/followees). This paper proposes a lightweight approach for the stance classification of users and keywords in OSNs, aiming at understanding the collective opinion of individuals and their concerns. Our approach employs a tailored random walk model, requiring just one keyword representing each stance, using solely the keywords in social media posts. Experimental results demonstrate the superior performance of our method compared to the baselines, excelling in stance classification of users and keywords, with a running time that, while not the fastest, remains competitive.