科研之友
微信
新浪微博
Facebook
分享链接
摘要
In view of the low efficiency of traditional collaborative filtering algorithm in personalized recommendation against the backdrop of mass data era, this paper, based on MapReduce parallel programming model, proposes a modified parallelization of ItemCF algorithm, which parallelizes the traditional ItemCF on Hadoop platform. This paper also elaborates the steps and details of the parallelization. This algorithm solves the problem of the operation of the ItemCF algorithm in large-scale data. The result of the experiment shows that the improved ItemCF algorithm has obviously better performance and speedup in personalized recommendation for commodity than the one realized on single node.
- 出版日期2017-5
- 单位郑州大学
