With the development of Internet, personalized recommendation has played an important role in human modern lives. Since the number of users’ data is always large-scale, traditional algorithms cannot effectively cope with e-commerce personalized recommendation tasks. This paper proposes an e-commerce product personalized recommendation system based on learning clustering representation. Traditional kNN method has limitation in selecting adjacent object set. Thus, we introduce neighbor factor and time function and leverage dynamic selection model to select the adjacent object set. We combine RNN as well as attention mechanism to design the e-commerce product recommendation system. Comprehensive experimental results have shown the effectiveness of our proposed method.

随着互联网的发展，个性化推荐已在人类现代生活中发挥了重要作用。由于用户数据的数量始终是大规模的，因此传统算法无法有效应对电子商务个性化推荐任务。本文提出了一种基于学习聚类表示的电子商务产品个性化推荐系统。传统的kNN方法在选择相邻对象集方面存在局限性。因此，我们引入了邻居因子和时间函数，并利用动态选择模型来选择相邻的对象集。我们结合RNN和注意力机制来设计电子商务产品推荐系统。综合实验结果表明了该方法的有效性。