Filter Bubbles and Polarization in News Recommendation: Evolution and Intervention
Abstract
Recent work in news recommendation has demonstrated that recommenders can over-expose users to articles that support their pre-existing opinions. However, most existing work focuses on a static setting or over a short-time window, leaving open questions about the long-term and dynamic impacts of news recommendations. In this thesis, we explore these dynamic impacts through a systematic study of three research questions: 1) How do the news reading behaviors of users change after repeated long-term interactions with recommenders? 2) What do different levels of vulnerability to recommendation influence have on this impact? 3) How to effectively alleviate such a procedure of polarization? Concretely, we conduct a comprehensive data-driven study through simulation experiments of political polarization in news recommendations based on 40,000 annotated news articles. We find that users are rapidly exposed to more extreme content as the recommender evolves and the inherent political preferences of users are getting increasingly radical. We also propose two complementary methods to dynamically alleviate such polarization, and we empirically show that combining these two methods together achieves a significant intervention effect.
Description
Keywords
filter bubble, recommender system, dynamic, calibration