一种基于链路优化的社团挖掘算法及其应用  

COMMUNITY DETECTION ALGORITHM BASED ON LINK OPTIMIZATIONAND ITS APPLICATION

作　　者：姚鑫宇 肖玉芝[1,2,3] 赵洪凯 Yao Xinyu;Xiao Yuzhi;Zhao Hongkai(College of Computer,Qinghai Normal University,Xining 810016,Qinghai,China;Tibetan Information Processing and Machine Translation Key Laboratory of Qinghai Province,Xining 810016,Qinghai,China;Key Laboratory of Tibetan Information Processing,Ministry of Education,Xining 810016,Qinghai,China)

机构地区：[1]青海师范大学计算机学院,青海西宁810016 [2]青海省藏文信息处理与机器翻译重点实验室,青海西宁810016 [3]藏文信息处理教育部重点实验室,青海西宁810016

出　　处：《计算机应用与软件》2023年第8期32-37,79,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(61763041,11661069);青海省新开科技计划项目(2020-GX-112)。

摘　　要：为了进一步优化Louvain算法在社团挖掘上的质量,使用节点相似度来确定社团初始形状,提出一种基于链路优化的P-L(Priority-Louvain)算法。与经典Louvain算法、GN算法和LPA算法在3个评价准则上进行对比。结果表明,该算法计算效果更好。实验构建了基于“用户-App”的两层网络,并对用户使用的App应用进行了社团划分,准确地划分出如爱奇艺、抖音和快手等App应用社团,可为通信行业服务商设计流量套餐提供策略参考信息。In order to further optimize the quality of Louvain algorithm in community mining,the node similarity was used to determine the initial shape of the community,and the P-L(Priority-Louvain)algorithm based on link optimization was proposed.The P-L algorithm was compared with the classic Louvain algorithm,the GN algorithm and the LPA algorithm on three evaluation criteria.The result shows that the calculation effect of P-L algorithm is better.We proposed and constructed a two-layer network on“user-App”,and the App was accurately divided into different community by the algorithm,such as iQIYI,Douyin and Kuaishou.The result of this study offered the reference for designing data package in the communications industry.

关 键 词：社团挖掘 优化链路 Priority-Louvain算法 通信网络 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]