基于加权小世界网络模型的用户聚类技术

User clustering technology based on weighted small world network model

传统的无加权小世界网络在用户聚类中具有良好的性质,但不能准确表达用户之间的紧密关系,导致用户聚类效果不够良好。为此在网络中引入了加权边,并利用小世界网络理论对系统中的用户网络进行分析,定义和计算了流动阻力和流动效率,建立了加权的小世界网络模型,并提出了此模型下的重连边算法。实验表明,与传统的小世界网络模型相比,该模型能更好地对用户进行聚类,收敛速度更快,聚类效果更好。

Traditional small word network has a good property in user clustering,but it is not an accurate expression of the close relationship between users,resulting in not good clustering effect of users.This paper introduced weighted edges in the network,analyzed user network in systems using small world network theory,defined and calculated the flow resistance and flow efficiency,built a user clustering model with small world properties,and proposed a re-connected edge algorithm.Experimental results show that,compared with the traditional small word model,the proposed model can be better for users to cluster,and also has faster convergence rate.It also achieves better clustering results than traditional clustering method.