基于FCM用户聚类的协同过滤推荐算法

Collaborative Filtering Recommendation Algorithm Based on Fuzzy C-Means User Clustering

传统的协同过滤推荐算法存在数据稀疏性以及推荐准确率低等问题,针对该问题提出一种基于模糊C均值聚类的协同过滤推荐算法GAFCM-CF(genetic algorithm based fuzzy c-means collaborative filtering)。首先,该算法结合用户评分和项目特征构建用户特征偏好矩阵,深入挖掘利用用户隐藏信息。其次,该算法通过模糊C均值聚类算法对用户进行聚类,并且为了防止模糊C均值聚类算法收敛于局部极小值,影响推荐质量,该算法基于遗传算法对模糊C均值聚类算法进行了改进,防止出现局部最优解。最后,该算法综合考虑了用户特征偏好矩阵以及用户项目评分矩阵计算用户相似度,实现推荐。实验结果表明,所提出的基于改进模糊C均值聚类的协同过滤推荐算法相比于传统的基于用户的协同过滤推荐算法及PDSFCM算法具有更好的推荐质量,提高了推荐的准确率。

Aiming at the problem of data sparsity and low accuracy of traditional collaborative filtering recommendation algorithms,a new genetic algorithm based fuzzy c-means collaborative filtering recommendation algorithm named GAFCM-CF is proposed.Firstly,the user feature preference matrix is constructed based on user rating matrix and item characteristics,and the hidden information of users is deeply mined.Secondly,the fuzzy c-means clustering algorithm is used to cluster the users.In order to prevent the fuzzy c-means clustering algorithm from converging to the local minimum and affecting the recommendation quality,the proposed algorithm improves the fuzzy c-means clustering algorithm based on genetic algorithm to prevent the local optimal solution.Finally,the user similarity is calculated by considering both the user characteristic preference matrix and user item rating matrix to realize better recommendation.The experiment shows that the proposed collaborative filtering recommendation algorithm based on improved FCM has better recommendation quality and improves the accuracy of recommendation compared with the traditional user-based collaborative filtering recommendation algorithms and PDSFCM algorithm.