面向海量数据的推荐系统的研究  

Research on recommendation system for mass data

作　　者：刘进[1] 胡大权[1] 陈家佳[1] 

机构地区：[1]重庆邮电大学,重庆400065

出　　处：《现代电子技术》2016年第12期59-61,65,共4页Modern Electronics Technique

基　　金：教育部人文社科规划项目:大数据时代海量电信客户价值评价方法研究(15XJA630003);重庆邮电大学社会科学基金项目(k2014-111)

摘　　要：在处理大数据时,传统的推荐系统,如常规协同过滤的推荐性能受到了限制。使用操作简便的K均值聚类算法与协同过滤构成组合推荐算法具有较好的推荐性能,该文使用遗传算法对组合推荐算法进行优化,简化组合推荐算法,降低组合算法的复杂度和成本。同时,通过对遗传算法进行改进,以提高遗传算法的优化能力,提高推荐系统性能。最后,通过MovieLens电影打分数据集对该文研究的推荐算法进行性能测试。结果表明,遗传算法的优化能力得到提升,推荐系统的性能有所提高。When dealing with large data,the traditional recommendation system,such as conventional collaborative filtering,is limited in its recommendation performance. The combination recommendation algorithm formed by K means clustering algorithm and collaborative filtering algorithm has better recommendation performance. The genetic algorithm is used in this paper to optimize the combination recommendation algorithm to simplify the algorithm so as to reduce the complexity and cost of the combination algorithm. At the same time,the genetic algorithm is ameliorated to improve the optimization ability of the genetic algorithm and the performance of the recommendation system. The performance test of the recommendation algorithm was carried out by means of the Movie Lens film scoring data set. The test results show that the optimization ability of the genetic algosithm and the performance of recommendation system have been improved.

关 键 词：大数据 推荐系统 协同过滤 遗传算法 K均值聚类 

分 类 号：TN911-34[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程]