Analysis of User Interests Based on Integrated-Graph Model

An integrated-graph model for user interests in personalized recommendation, which is based on small-world network and Bayesian network, is presented. The integrated-graph model consists of two layers. One is user´s layer for representing users or customers and the other is merchandise´s layer for representing goods or produce. The relationships among users are described by small-world network at lower layer. The implications among merchandises are represented by Bayesian network at higher layer. Directed arcs denote the interests and tendency between user´s layer and merchandise´s layer. Several algorithms for clustering and interest analysis based on small-world network are introduced. The experimentation shows that the model can well represent the relationships among users to users, merchandise to merchandise, and users to merchandise. The result of interest recommendation based this integrated-graph mode is better than others.