基于模糊C-均值和特征加权法的协同过滤推荐算法

Collaborative filtering algorithm based on fuzzy C-means and weighted method of eigenvalues

随着用户对推荐精准度和个性化的要求越来越高,传统推荐算法对于用户-项目的非理性评判弊端以及稀疏性等带来的问题,严重影响推荐的精准度及个性化。基于这些问题,提出了一种基于C-均值和特征加权法的协同过滤推荐算法,该算法首先对用户-项目的评分、转存、特征值等建立矩阵来计算它们之间的相似度,然后通过FCM算法筛选出候选集;而候选集中的项目、用户又具有特征值权重向量的划分,所以再对候选集通过特征加权法获得最终推荐集,从而实现精准、个性化及高效推荐。实验结果表明,该算法能够有效提高相似度计算精度,从而解决数据稀疏性及评分极端带来的不良性,进而提高推荐的精准度、个性化及执行效率。

With the increasing demands of users for the precision and personalization of recommendation,the shortages of traditional recommendation algorithm have a serious impact on the accuracy and personalization of the user-project＇s irrational judgment and sparsity. Based on these problems,in this paper,a collaborative filtering recommendation algorithm based on C-means and feature weighting is proposed,which firstly calculates the similarity between the user-project＇s scoring,the dump,the eigenvalue and so on,and then filters out the candidate set by FCM algorithm. The candidate concentration project and the user also have the characteristic value weight vector＇s division,therefore the final recommendation set can beobtained through the characteristic weighting method to the candidate set,thus the accurate,personalized and highly effective recommendation are realized. The experimental results show that the algorithm can effectively improve the accuracy of similarity calculation to solve the problem of data sparsity and grading,and improve the precision,individuation and execution efficiency of the proposed method.