摘要
传统社区挖掘往往不具灵活性,并且忽略用户节点特性.为能更加个性化地挖掘网络中的社区,在传统的社交网络表示方法的基础上,提取用户在社交网络中的属性,例如用户档案中的信息、用户的度节点信息等.将得到的数据进行结构化分类处理,用布尔值表示静态属性,用连续值表示动态属性,再将两者相结合构建混合型数据的贝叶斯网络.并通过图论简化用户网络信息结构,优化计算过程,最后对模型进行可行性检验.实验最后结果表明社区挖掘具有较高的精确性,且更具灵活性,能应用于各种社区的挖掘.
Traditional community mining in social network always neglected users' attributes. However, the research found that some information are important factors for social community division, such as user profile, out-degree and in-degree. For this reason, based on the traditional representation of social net- work, this paper extracted network users~ attributes and analyzed the current users~ profiles in social net- work. Then users~ data are classified into two categories, one is static attribute comprised by Boolean val- ue; the other is dynamic data which consists of continuous value. After this, according to the relationship between the nodes, identified the corresponding sub nodes and the parent nodes then build up the structure of Bayesian network of Hybrid data. Finally, the graph-theory algorithm is used to simplify its calculation and test the precision and accuracy of the model. This model can provide rapid, accurate, and personalized technology of community mining with users~ attributes.
出处
《武汉大学学报(工学版)》
CAS
CSCD
北大核心
2014年第3期426-432,共7页
Engineering Journal of Wuhan University
基金
国家自然科学基金项目(编号:70971027)
关键词
用户属性
社区挖掘
贝叶斯网络
图论
社交网络
users
attributes
community mining
Bayesian networks
graph theory
social networks