Online Recommendation based on Collaborative Topic Modeling and Item Diversity

Abstract

Online news websites provide diverse article topics, such as fashion news, entertainment and movie articles to attract more users and create more benefits. Analyzing users' browsing behaviors and preferences to provide online recommendations is an important trend for online news websites. In this research, we propose a novel online recommendation method for recommending movie articles to users when they are browsing the news. Specifically, association rule mining is conducted on user browsing news and movies to find the latent associations between news and movies. A novel online recommendation approach is proposed based on Latent Dirichlet Allocation, enhanced Collaborative Topic Modeling and the diversity of recommendations. We evaluate the proposed approach via an online evaluation on a real news website. The online evaluation results show that our proposed approach can enhance the click-through rate for recommending movie articles and alleviate the cold-start issue.