Improvement of Item-Based Collaborative Filtering by Adding Time Factor and Covering Degree
Zhipeng Zhang, Yasuo Kudo, Tetsuya Murai · 2016
Item-based collaborative filtering (IBCF) is an important technology that is widely used in recommender system. It uses historical information to compute item-item similarity and make rating predictions. However, current IBCF approaches have a problem in that all items are accorded the same weight when computing the similarity and making predictions. To improve the quality of recommendations made by IBCF, we considered the memory habits of a customer and applied the covering-based rough set theory. In this research, we introduced a time-based correlation degree and applied it to the computation of item-item similarity, the covering degree was applied to make rating prediction. Our experimental results suggest that this novel approach produces recommendation results superior to those of existing work.