基于离群数据挖掘的电子商务推荐系统研究

Research on Recommender System of e-commerce based on Outlier Mining

电子商务推荐系统在推荐的精度和实时性方面,往往存在冲突,即为了提高实时性,会造成推荐精度不高;为提高推荐的整体质量,造成实时性不够准确。对此,找到推荐精度和实时性之间的契合点,是提高系统推荐的重点。本文结合离群数据的特点,提出一种基于改进K均值和PSO的混合算法。针对传统K均值算法在对离群数据挖掘中存在不足的基础上,引入PSO算法,并对欧氏距离、学习因子聚类流程等进行改进,然后对聚类推荐算法流程进行改进。最后,通过试验对比,验证了改进算法在实时性和准确性方面都有较大的提升,由此证明混合算法的可行性。

the electronic commerce recommendation system in the recommendation accuracy and real time, there is often conflict, namely,in order to improve the real-time,will cause the recommendation accuracy is not high;in order to improve the overall quality of the resulting real time is not accurate enough. In order to improve the system recommendation, it is important to find the conjunction between recommendation accuracy and real-time performance. Based on the characteristics of outlier data, a hybrid algorithm based on im- proved K mean and PSO is proposed in this paper. In view of the traditional K-means algorithm in K of ouflierdetection based on the lack of PSO algorithm is introduced, and the Euclidean distance, the clustering process of learning factor is improved, then the improved clustering recommendation algorithm process. Finally, the experimental results show that the improved algorithm is more efficient in real- time and accuracy, and the feasibility of the hybrid algorithm is proved.