摘要
为解决异构信息网络相似性度量的问题,提出了基于节点特征子图的节点相似性度量算法,通过节点特征子图的最大公共子图与最小公共超图之间的差异性,进行节点间的相似性度量。该算法以图理论为基础,根据连边的不同类型设定不同权值,在考虑节点信息相似的同时,加入节点在网络中的结构信息,最大程度地利用了异构信息网络所富含的信息。实验结果表明,提出的算法具有较好的性能和有效性。
To solve the problem in measuring the similarity of heterogeneous information networks, a similarity measuring algorithm was proposed. It calculates the difference between the maximum common sub-graph and minimum common hyper-graph, based on feature sub-graph of the current node. The algorithm takes graph theory as its foundation, set different weight to different kinds of edges, considers nodes information as well as graph to topological information, and makes full use of the information in heterogeneous network. The result shows that the proposed algorithm has wonderful effectiveness and efficiency.
出处
《电信科学》
北大核心
2014年第11期66-72,共7页
Telecommunications Science
基金
国家自然科学基金资助项目(No.61103043
No.61173099
No.U1233118)
国家"十二五"科技支撑计划基金资助项目(No.2012BAG04B0)
武汉大学软件工程国家重点实验室开放基金资助项目(No.SKLSE2012-09-26)
关键词
异构信息网络
图相似
相似性度量
特征子图
heterogeneous information network, graph similarity, similarity measurement, feature sub-graph