Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (5): 118-123.DOI: 10.3778/j.issn.1002-8331.1804-0318

Previous Articles     Next Articles

Collaborative Filtering Recommendation Algorithm Based on Improved Fuzzy Partition Clustering

SU Qing1, ZHANG Jingfang1, LIN Zhengxin1, LI Xiaomei1, CAI Zhaoquan2, ZENG Yong’an1   

  1. 1.School of Computer, Guangdong University of Technology, Guangzhou 510006, China
    2.Department of Computer Science and Technology, Huizhou University, Huizhou, Guangdong 516007, China
  • Online:2019-03-01 Published:2019-03-06

改进模糊划分聚类的协同过滤推荐算法

苏  庆1,章静芳1,林正鑫1,李小妹1,蔡昭权2,曾永安1   

  1. 1.广东工业大学 计算机学院,广州 510006
    2.惠州学院 计算机科学与技术系,广东 惠州 516007

Abstract: The traditional Collaborative Filtering(CF) recommendation algorithm has the defects of sparse score matrix, weak extensibility and low recommendation accuracy. A collaborative filtering recommendation algorithm(GIFP-CCF+) is proposed to improve the fuzzy partition clustering. In the traditional calculation method based on modified cosine similarity, the time-difference factor, hot-item weight factor and cold-item weight factor are introduced to improve the similarity calculation results. At the same time, the GIFP-FCM algorithm which improves the fuzzy partition is introduced to form a class of items with similar attributes, construct index matrix, and base on the index. The similarity between items finds the nearest neighbor recommendation of the project, thereby improving the accuracy of the Collaborative Filtering algorithm(CF). By comparing with Kmeans-CF, FCM-CF and GIFP-CCF algorithms, it is proved that the GIFP-CCF+ algorithm has some advantages in recommendation result and recommendation precision.

Key words: recommender system, collaborative filtering, improved fuzzy partition, fuzzy C means clustering

摘要: 针对传统协同过滤(CF)推荐算法存在评分矩阵稀疏、扩展性弱和推荐准确率低的缺陷,提出一种改进模糊划分聚类的协同过滤推荐算法(GIFP-CCF+)。在传统基于修正余弦相似度计算方法上,引入时间差因子、热门物品权重因子以及冷门物品权重因子以改善相似度计算结果;同时引入改进模糊划分的GIFP-FCM算法,将属性特征相似的项目聚成一类,构造索引矩阵,同索引间根据项目间的相似度寻找项目最近邻居构成推荐,从而提高协同过滤算法(CF)的精度。通过与Kmeans-CF、FCM-CF和GIFP-CCF算法进行仿真对比实验,证明了GIFP-CCF+算法在推荐结果和推荐精度上具有一定的优越性。

关键词: 推荐系统技术, 协同过滤, 改进模糊划分, 模糊[C]均值聚类