基于社区与社交关系的跨网络身份匹配模型
A Cross Network Identity Matching Model Based on Community and Social Relationships
马廷淮 1黄锐 1彭可兴 1荣欢2
作者信息
- 1. 南京信息工程大学计算机学院,江苏 南京 210044
- 2. 南京信息工程大学人工智能学院,江苏 南京 210044
- 折叠
摘要
跨社交网络身份匹配技术可以判断来自不同社交网络的账号是否属于同一个自然人.但社交网络拓扑结构复杂,数据庞大,因此匹配准确的同时节省计算资源与时间仍然是一个巨大的挑战.针对上述问题,提出基于社区结构与社交关系的跨社交网络身份匹配方法(IMCS).IMCS利用社交结构与跨网络对齐的互动值向量进行用户表征,解决用户跨网络社交数据分布具有差异性的问题,并通过划分跨网络社区结构提出优先匹配机制,减少无效匹配次数并提高匹配的精度.基于真实社交网络数据集的实验结果表明,IMCS的表现优于对比算法,并在提升跨社交网络身份匹配准确率的同时降低无效匹配次数.
Abstract
User identity linkage across social networks can determine whether different accounts from diverse social networks belong to the same identity.However,the complex topology of social networks and the huge amount of data make it a huge challenge to match accurately while saving computational resources and time.To address these problems,a cross-social network identity matching method based on community structure and social relationships(IMCS)is proposed.IMCS uses social structures and cross-network aligned interaction vectors for user representation to tackle the challenge of disparities in the distribution of user data across social networks,and we propose a priority matching mechanism by dividing the cross-network community structure to reduce the number of invalid matches and improve the accuracy of matching.Experimental evaluations are conducted based on real social network datasets dem-onstrate that IMCS outperforms the comparison methods and reduces the number of invalid matches while improving the cross-network identity matching accuracy.
关键词
社交网络/用户匹配/跨网络/网络嵌入/社区结构Key words
Socialnetworks/User matching/Cross-social network/Network embedding/Community structure引用本文复制引用
基金项目
国家重点研发计划(2021YFE0104400)
国家自然科学基金(62102187)
出版年
2024
NETL
NSTL
万方数据