国外电影推荐系统网站研究与评述  

Review on dominating websites for movie recommender systems

作　　者：孙海峰[1] 甘明鑫[1] 刘鑫[1] 吴越[1] 

机构地区：[1]北京科技大学东凌经济管理学院,北京100083

出　　处：《计算机应用》2013年第A02期119-124,共6页journal of Computer Applications

基　　金：国家自然科学基金资助项目(71101010);中央高校基本科研业务费资助项目(FRF-BR-11-019A)

摘　　要：电影推荐网站的研究对于解决电影数据复杂性以及用户需求多样性具有重要的意义。针对电影推荐系统网站中推荐准确率低、数据稀疏和冷启动等问题,详细阐述了国外六大主流电影推荐网站系统的内部结构、信息量和推荐方式;重点阐述了各网站的推荐异同以及可优化方案,进而通过新用户对电影进行评分的冷启动实验,详细分析冷启动解决方法与效果;最后指出当前电影推荐网站的现状及问题,对推荐系统相关理论及其发展趋势进行了分析和展望,并为实际应用中解决冷启动与数据稀疏等问题提供参考依据。The research and practice d movie recommender system is important in solving the complexity of movie data and diversity of user requirements. Focused on the problems of low accuracy, data sparsity and cold start in six mainstream foreign movie-recommend websites, this paper summarized the systematic internal structure, the detailed recommended way, the movie information, etc. This paper also emphazed on the similarities and differences of the recommendation ways among the websites and the optimized proposal; and furthermore, according to the grading of movies by new users, the cold start experiment analyzed the way and effect of cold start solutions. Finally, the paper clearly indicated the situation of current movie recommendation websites. The opening issues and challenges for the movie recommendation system concerning both the theory and trend were proposed, which provide a reference for practical application in the solution of cold start and data sparsity.

关 键 词：电影推荐系统 协同过滤 冷启动 数据稀疏 标签 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]