Research On User Modeling And Community Partition Algorithms

Personalized service is an effective method in solving the "data explosion" and "information trek". As the core technologies of personalized services, user modeling and community division have recently attracted much attention. The quality of the personalized services is determined by the accurancy of user modeling and community division. Therefore, in order to improve the quality of personalized services, a user model correction algorithm based on assimilation effect is proposed by amending user model with the consideration of assimilation effect, and a community partition algorithm based on abstract distance and improved modularity is proposed in this paper.First, the need to consider assimilation effect in user modeling is raised with the analysis of the process of user modeling, and the assimilation effect function is proposed by analysing the generation and influence factors of assimilation effect. Second, suitable community partition measures and standards for social network: abstract distance and improved modularity are put forward by analysing the characteristics of social network. And a social network community partition algorithm is proposed based on abstract distance and improved modularity.Finally, the proposed algorithms are verified on the Matlab platform. Experiments show that the accuracy of user model proposed in this paper is 78%. The NMI and modularity show the community partition algorithm proposed in this paper has better performance. So taking the influence of assimilation effect into account in user modeling can improve the accuracy of user modeling and the community partition algorithm proposed in this paper can highly improve the values of NMI and modularity.