How News Recommendation System Effects User’s Individual and Aggregate Topic Diversity—A Study of Simulation Analysis

This paper is to explore the influence of recommendation system on users' topic diversity. We construct a simulation system that covers the news recommendation closed-loop processes of user profile generation, news generation and delivery, user selection and browsing, and user profile update. Quantitative indicators were formulated to assess users' topic diversity, encompassing both individual and aggregate dimensions. Then simulation compared the effects on users' topic diversity across three news delivery methods. The results show that compared with the delivery method of self-selection, filtering-based recommendation and popular recommendation significantly increase users' individual topic diversity while having obstructive effects on users' aggregate topic diversity. Various recommendation algorithms and ways to update user profiles have slight impacts and do not change relationships in trends. Lastly, we discussed the relevant topics in conjunction with our conclusions.