基于结构平衡理论和高阶互信息的符号网络表示学习算法  

Signed Network Representation Learning Algorithm Based on Structural Balance Theory and High-Order Mutual Information

作　　者：郁湧[1,2] 钱天宇 高悦 艾合买提尼牙孜 刘金卓[1,2] YU Yong;QIAN Tianyu;GAO Yue;AIHEMAITINIYAZI;LIU Jinzhuo(School of Software,Yunnan University,Kunming 650504;Key Laboratory in Software Engineering of Yunnan Province,Yunnan University,Kunming 650504;Engineering Research Center of Cyberspace,Ministry of Education,Yunnan University,Kunming 650504)

机构地区：[1]云南大学软件学院,昆明650504 [2]云南大学云南省软件工程重点实验室,昆明650504 [3]云南大学跨境网络空间安全教育部工程研究中心,昆明650504

出　　处：《电子科技大学学报》2023年第5期780-788,共9页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金(61976181);云南省科技厅面上项目(202001BB050063);云南省软件工程重点实验室开放项目(2020SE315)。

摘　　要：提出了一种基于结构平衡理论和高阶互信息的符号网络表示算法SNSH,通过反转符号网络中的正负关系生成负图,来挖掘符号网络中隐含的高阶互信息。该方法旨在通过加强的社会平衡理论来模拟符号网络的局部隐含特征,并通过节点局部嵌入、网络全局结构和节点特征属性三者之间的高阶互信息,得到更全面的符合符号网络特性的节点嵌入。In this paper,a symbolic network representation learning framework signed network representation learning algorithm based on structural balance theory and high-order mutual information(SNSH)is proposed.By reversing the positive and negative relationships in symbolic networks to generate negative graphs,the hidden high-order mutual information in symbolic networks is mined.This method aims to simulate the local implicit features of symbolic networks through the strengthened social balance theory,and obtain a more comprehensive node embedding that conforms to the characteristics of symbolic networks through the high-order mutual information among the local embedding of nodes,the global structure of the network and the characteristic attributes of nodes.

关 键 词：高阶互信息 链路符号预测 网络表示学习 符号网络 结构平衡理论 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]