2014/05/29 by Nguyễn Thanh Hải, Jérémie Mary, Nguyen, Hai Thanh +3 · 1 citation
Computer Science · Decision Sciences · #Advanced Bandit Algorithms Research #Data Stream Mining Techniques #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Optimization and Search Problems #Recommender Systems and Techniques
paper · pdf · doi:10.48550/arxiv.1405.7544
openalex publication_date 2014/05/29 · openalex created_date 2022/10/05 · openalex updated_date 2026/07/28
In this paper, we study a cold-start problem in recommendation systems where\nwe have completely new users entered the systems. There is not any interaction\nor feedback of the new users with the systems previoustly, thus no ratings are\navailable. Trivial approaches are to select ramdom items or the most popular\nones to recommend to the new users. However, these methods perform poorly in\nmany case. In this research, we provide a new look of this cold-start problem\nin recommendation systems. In fact, we cast this cold-start problem as a\ncontextual-bandit problem. No additional information on new users and new items\nis needed. We consider all the past ratings of previous users as contextual\ninformation to be integrated into the recommendation framework. To solve this\ntype of the cold-start problems, we propose a new efficient method which is\nbased on the LinUCB algorithm for contextual-bandit problems. The experiments\nwere conducted on three different publicly-available data sets, namely\nMovielens, Netflix and Yahoo!Music. The new proposed methods were also compared\nwith other state-of-the-art techniques. Experiments showed that our new method\nsignificantly improves upon all these methods.\n