弱连接视角下的高影响力节点识别  

High-influence Node Recognition from the Perspective of Weak Ties

作　　者：姚月娇 刘向[1] 余博文 Yao Yuejiao;Liu Xiang;Yu Bowen(School of Information Management,Central China Normal University,Wuhan 430079,China)

机构地区：[1]华中师范大学信息管理学院,湖北武汉430079

出　　处：《现代情报》2022年第9期58-67,共10页Journal of Modern Information

基　　金：国家自然科学基金项目“专利引证网络中创新节点的浮现与长期演化研究”(项目编号:71671306)。

摘　　要：[目的/意义]PageRank是被普遍接受并广泛使用的排序算法,通过在节点之间传递PR值识别网络中的重要节点。针对PageRank算法对节点间连接强度和影响强度的忽视问题,提出TransRank算法以更有效地挖掘高影响力节点。[方法/过程]TransRank算法利用节点相似性度量连接强度,再基于弱连接理论拟合连接强度和影响强度的关系,并将影响强度作为传值依据。[结果/结论]通过在3个现实网络中进行SI传染实验,检验TransRank算法的优化效果,结果显示TransRank算法挖掘出的高影响力节点在影响速度上始终优于PageRank算法,在影响范围上有很大可能优于PageRank算法。[Purpose/Significance]PageRank is a generally accepted and widely used ranking algorithm, which identifies important nodes in the network by passing PR values between nodes.Aiming at the problem of the PageRank algorithm’s neglect of the connection strength and influence strength between nodes, the TransRank algorithm is proposed to mine high-impact nodes more effectively.[Method/Process]The TransRank algorithm used node similarity to measure the connection strength, and then matched the relationship between the connection strength and the influence strength based on the weak tie theory, and used the influence strength as the basis for value transmission.[Result/Conclusion]Through SI infection experiments in three real networks, the optimization effect of the TransRank algorithm is tested.The results show that the high-impact nodes mined by the TransRank algorithm are always better than the PageRank algorithm in terms of impact speed, and are likely to be better than the PageRank algorithm in terms of impact range.

关 键 词：弱连接 节点相似性 连接强度 影响强度 节点影响力 

分 类 号：G250.252[文化科学—图书馆学]