物理学报Issue(5)：1-9,9.DOI:10.7498/aps.64.058902
面向结构洞的复杂网络关键节点排序∗
Ranking key no des in complex networks by considering structural holes
摘要
Abstract
Structural hole nodes in complex networks play important roles in the network information diffusion. Unfortunately, most of the existing methods of ranking key nodes do not integrate structural hole nodes and other key nodes. According to the relevant research on structural hole theory as well as the key node ranking methods, network constraint coefficient, betweenness centrality, hierarchy, efficiently, network size, PageRank and clustering coefficient, 7 metrics are selected to rank the key nodes. Based on the 7 metrics, a ranking learning method based on ListNet is introduced to solve ranking key nodes by multi metrics. Comprehensive experiments are conducted based on different artificial networks and real complex networks. Experimental results with manual annotation show that the ranking method can comprehensively consider the structural hole nodes and other nodes with different important features. The ranking results on different networks are highly consistent with the manual ranking results. The spreading experiment results using signed to interference ratio propagation model show that SIR model can reach a maximum propagating ratio in a shorter propagating time initiated by TOP-K key nodes selected by our method than TOP-K key nodes selected by other methods.关键词
结构洞/关键节点/排序学习/复杂网络Key words
structural hole/key node/learning ranking/complex network引用本文复制引用
韩忠明,吴杨,谭旭升,段大高,杨伟杰..面向结构洞的复杂网络关键节点排序∗[J].物理学报,2015,(5):1-9,9.基金项目
国家自然科学基金(批准号:61170112)、中央财政支持地方高校发展专项资金人才培养和创新团队建设项目(批准号:19005323132)、教育部人文社会科学研究基金项目(批准号:13YJC860006)资助的课题.@@@@* Project supported by the National Natural Science Foundation of China (Grant No.61170112), the central finance special fund to support the development of local colleges and Universities, China (Grant No.19005323132), and the of Humanities and Social Science Research Fund Ministry of Education, China(Grant Nos 13YJC860006) （批准号:61170112）
