Comprehensive Temporal Filter for Expanded Collaborative Filtering Algorithm

Abstract

Collaborative filtering-based recommendation has been introduced in e-commerce systems, and it can save time and cost for retrieval large scale online items. However because traditional collaborative filtering algorithm depends on item rating scores, it is hard for recommendation systems to reflect temporal properties. In order to improve this problem, temporal decay filter has been proposed, and it mainly focuses on the elapse time of individual item after the initial rating time rather than general temporal similarity among the entire items rated in common by multiple users. From this motivation, this paper proposed the comprehensive temporal filter so that global temporal similarities of common items are considered by the recommendation algorithm, and the performance has been evaluated by the comparison with existing methods.