关键词: 混合推荐算法, 增量学习, 用户建模, 基于内容的过滤, 协作过滤

Abstract: Recommendation is a kind of personalized service in the adaptive information system, and it can provide personalized information according to individual information needs. Concerning the issues of the user profiling granularity and the sparsity of user-item matrix in the classical collaborative filtering, a hybrid recommendation approach named WHHR was proposed based on incremental learning. WHHR implemented user modeling with Widrow-Hoff learning and predicted ratings by collaborative filtering. The experimental results demonstrate that the proposed approach outperforms other similar ones in convergence and accuracy.

Key words: hybrid recommender approach, incremental learning, user profiling, content-based filtering, collaborative filtering