The study of users' relationship community discovering and on that basis the users' preference modeling is of important significance. Based on the elaboration of mechanism and research advance of community discovery in complex network, this article improved the discovering method of relationship community from the perspective of modified modularity according to the diversification of users' preference, the small-scale and overlapping of users' relationship community. And then, the users' preference model was constructed from two levels: community and individual. A new proposal was put forward that the weight fusion of community model and individual model was made to achieve the users' whole model. The comparative experiment demonstrated that this method is more superior to traditional methods from the perspective of recall ratio.%本文着重研究了社会网络环境下的用户关系社区发现及在此基础上的用户兴趣建模问题.在阐述复杂网络中社区发现机理和研究进展的基础上,本文针对社会网络环境下用户兴趣多元化及关系社区小规模化和交叉性等特点,从模块度改进的角度进行关系社区发现算法的改进.进而从社区和个体两个层面进行了用户兴趣模型构建,提出将两者加权融合实现用户整体建模的思路.对比试验表明,基于关系社区的用户建模在在查全率方面具有优越性.
展开▼