Deep Recommender Systems Utilizing Side Information

ABSTRACTRecommendation Systems (RS) are designed to assist users in decision making by recommending the most appropriate information or products for them. Nonetheless, many RS suffer from limitations such as data sparsity and cold-start. Side information (SI) can be integrated into a recommender system to tackle these limitations. In my Ph.D. research, I seek to build on and extend the use of SI for RS. Specifically, I propose new types and representations of SI and develop new methods to integrate SI into RS to boost its performance. This paper presents the conceptual foundation and motivation of my Ph.D. research.