Understanding longitudinal dynamics of recommender systems performance: An agent-based modeling approach

We develop a general agent-based modeling and simulation approach to study the impact of various factors on the temporal dynamics of recommender systems' performance. In this paper, we specifically focus on exploring the product consumption strategies and show that, in the long run, they can significantly impact the system's recommendation quality in various ways. For example, the more heavily users rely on the system's recommendations to make item choices, the more inaccurate and less diverse (and, thus, arguably less useful) system's predictions become in the future, which has important implications for recommender systems design. More generally, the proposed agent-based simulation approach allows for comprehensive analysis of longitudinal recommender systems performance under a variety of diverse conditions, which typically is not feasible with live real-world systems.