Characterization of User Engagement in Electronic News Media: A Case Study for India
Proceedings of the Second Workshop on Building Educational Applications Using NLP
Abstract
Online news platforms have become central spaces for political discourse, playing a critical role in shaping public opinion and democratic participation. In large-scale democracies such as India, the combination of extensive user engagement, ideological polarization, and automated participation raises significant concerns regarding trust, transparency, and manipulation. This paper presents a large-scale empirical study of political discourse on a prominent Indian news platform, analyzing over 21,000 news articles and more than 1.5 million user comments. We investigate how ideological bias, sentiment dynamics, and non-organic user behavior interact to shape engagement patterns. Our methodology integrates hybrid article bias classification, large-scale sentiment analysis, heuristic-based bot detection, coordinated behavior analysis, and a focused examination of super-active users. In addition, we compare rule-based stance inference with large language model (LLM)-based stance classification to assess trade-offs between computational efficiency and contextual accuracy. The results reveal systematic sentiment skew, disproportionate influence by super-active and bot-like users, and coordinated campaigns aligned with specific political narratives. We conclude by discussing the implications of these findings for trust in online political discourse and reflecting on the dual role of generative AI as both an analytical tool and a potential vector for manipulation.