针对传统关联规则挖掘算法中频繁信息不完善,以及电子商务中各个影响因素贡献值不同的问题,提出一种基于矩阵多源加权关联个性化推荐模型——MSWPR模型,并在考虑虚拟行为水平加权和多源关联垂直加权的基础上,引入最小支持数概念作为剪枝的依据,进一步结合该模型对个性化推荐流程进行了概述。

In view of the traditional association rule mining algorithm of frequent information is imperfect,and the contribution of various factors in the e-commerce is different,this paper put forward a personalized recommendation model based on matrix multi-source weighted association---MSWPR model.And considering the virtual behavior level weighted and multi-source vertical weighted on the basis of correlation,a concept of minmum support count is introduced as the basis of pruning.Further combining with the model has summarized the personalized recommendation process.

国家自然科学基金资助项目：云环境用户多兴趣图谱的移动商务关联性推荐模型及算法研究，71271186；河北省自然科学基金项目：面向云计算的移动商务用户多源兴趣进化模型及关联性推荐算法研究，G2013203237