Evaluation of Tasks in Recommendation Systems
Chenglong’s research focuses on the evaluation principles of recommender algorithms. The research proposes a framework for recommender algorithms that can simulate user needs evolution and provide multi-faceted evaluation metrics. The framework is based on interdisciplinary theories such as social psychology and decision-making theory. According to the modeling of user behavior, the research proposes the principle and solution of how to generate recommendations that balance accuracy and diversity when population-scale interest drift occurs in a non-stationary environment.
Prof Mark Sanderson, RMIT University
Dr Yongil Ren, RMIT University
Assoc Prof Pablo Castells, Universidad Autónoma de Madrid