摘要
为了解决知识服务站点上传统知识分类模式的限制,根据社会学中集群行为观点,将网络用户的访问行为看做集群行为,且根据用户集群行为在知识服务网上对知识项的访问提出知识集群概念。结合访问日志库中的记录以及用户集群行为的时段性特点,利用贝叶斯网中变量之间概率依赖关系的优点,构建一种动态的、基于用户集群行为的知识聚合模型——知识集群,此模型将用户访问的知识项看做网络节点,利用概率推理得出节点之间的依赖关系,最终动态形成知识项聚合。通过实验数据,证明了此模型方法的可行性及有效性。
To address the limitation of traditional knowledge classification model in knowledge service website,presented a new concept,called knowledge clusters,based on the on-line access to the item of knowledge according to the view of clustering behavior in sociology,for seeing the user＇s access behavior of knowledge as clustering behavior.The paper built a dynamic model of knowledge aggregation,called knowledge cluster model,by regarding accessed item of knowledge as nodes in the Bayesian network and describing the dependencies between the nodes using the probabilistic reasoning method,and then to form the knowledge aggregation dynamically.At last,this model approach is proved to be feasible and valid through the expe-rimental data.
出处
《计算机应用研究》
CSCD
北大核心
2011年第3期828-830,844,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(70871043
70801028)
关键词
知识集群
贝叶斯网
知识聚合
知识项
knowledge cluster
Bayesian network
knowledge aggregation
knowledge item
