协同过滤推荐算法比较研究

Comparable Research on Collaborative Filtering Recommendation Algorithms

推荐系统被广泛用于电子商务、视频网站、新闻小说推荐等各个领域,旨在向用户提供其可能感兴趣的信息。协同过滤是推荐系统的主流技术,分为基于内存的协同过滤、基于模型的协同过滤以及组合协同过滤。以其中基于用户(项目)、基于矩阵分解以及基于线性回归集成策略的协同过滤算法为例进行说明比较,然后通过Movie Lens的数据集进行实验对比。

Recommender system designed to provide users with information that may be of interest is widely used in E-Commerce, video websites,news and novels recommendation, etc. Collaborative filtering is the main technology of recommender system, is divided into three classes：collaborative filtering based on memory, collaborative filtering based on the model and hybrid recommendation. Compares the algorithms of user（item） based, matrix factor based and linear regression in integration strategy as an example and gets the result through experimen-tal comparison with the dataset of Movie Lens system.