分布式矩阵分解算法在推荐系统中的研究与应用

The Research and Application of Distributed Matrix Factorization Algorithm in Recommend System

随着互联网信息技术的高速发展,越来越多的人开始接触网络。网上购物成为当今社会的购物的主要方式之一,各大电子商务网站每天都有大量的消费者购买及浏览记录。电子商业往往希望通过分析大量的网站购买记录以及消费者浏览记录,对消费者提供有价值地商品推荐,以便于提高销售量。矩阵分解广泛地应用在推荐系统中的协同过滤算法中,但是,随着网站数据量的增大,传统的矩阵分解算法不能有效地处理这些大规模海量数据。本文针对推荐系统中大规模的网站数据,提出了基于云平台的分布式矩阵分解算法,该算法能够分布式完成推荐系统中的协同过滤过程。实验结果表明,本文提出的算法能够高效地完成推荐系统中的协同过滤,并且,该算法具有很好的可扩展性。

With the high-speed development of interact information technology, more and more people begin to get in touch with internet. On-line shopping is becoming one of the main shopping methods, several big e-commerce sites produce big amount of consumer purchasing and browsing records. E-commerce sites usually hope to analyze these records and provide to the consumers with valuable commodity recommendation, in order to promote sales. Matrix factorization is popularly used in collaborative filtering algorithm in recommendation system. However, with the incensement of web data, traditional matrix factorization could not deal with this huge amount of data. In this paper, focusing on big scale website data, we propose distributed matrix faetorization based on Cloud computing platform. Through the experimental results, the algorithm in this paper could complete collaborative filtering in recommendation system effectively, and it has good scalability.