一种基于特征属性的Web用户模糊聚类改进算法

为降低传统FCM算法的计算复杂性，提高Web用户聚类的效果，文中提出了一种改进的基于特征属性的Web用户模糊聚类算法。首先通过用户访问页面的次数和时间建立Web用户兴趣度矩阵，并根据商品的特征属性值将Web用户兴趣度矩阵映射为用户对特征属性的偏好矩阵，从而有效降低数据稀疏性；然后以此为数据集，对传统的FCM算法进行了改进，将聚类中心分为活动和稳定两种，忽略稳定聚类中的距离计算以降低计算复杂性。最后通过仿真实验证实了新算法的有效性和可行性。

An Improved Web Users Fuzzy Clustering Algorithm Based on Features Property

In this paper, present an improved Web users fuzzy clustering algorithm based on features property to reduce the computational complexity of conventional fuzzy c-means clusering algorithm and improve the effect of Web users clusering. First,establish Web user＇s interest degree matrix through the times and time of user＇s visited pages, mapping the Web user＇s interest degree matrix into the user＇s fea- tures property preference matrix according to the features property of item to reduce the data sparseness effectively. Based on the features property preference matrix, improved the conventional fuzzy c-means clusering algorithm. The proposed method first classifies cluster centers into active and stable groups,then skips the distance calculations for stable clusters in the iterative process to reduce the computa- tional complexity of conventional fuzzy c-means clustering algorithm. Finally ,the simulation demonstrates the feasibility and validity of the proposed method.