摘要
[目的/意义]挖掘潜在好友关系并进行精准的好友推荐服务,已成为社交网络领域研究的热点,基于用户属性-关系相似度的好友推荐模型研究旨在增强用户忠诚度以及在线社区活跃度,提升社区的信息服务准确性和效率。[方法/过程]通过融合用户链接关系与属性特征,提出用户属性-关系相似评价体系;采用因子分析法,计算得出各项目权重以及综合得分;据此构建社交网络相似度矩阵,基于派系划分方法,对用户进行划分分区,最终实现好友推荐服务。[结果/结论]实验结果表明,运用派系划分的基于用户属性-关系推荐模型在推荐列表长度受限情况下的整体表现较优,有效提高推荐精准度。
[Purpose/significance]The specialists in the field of social networking have turned the spotlight on the issue of mining potential friends’relation to provide accurate friend recommendation service.The research on friend recommendation model based on user attribute-relation similarity is tend to enhance user loyalty and online community vibrancy and,in the meantime,the accuracy and efficiency of community information services.[Method/process]By combining user connections and characteristics,this paper firstly proposes the evaluation system of user relationship-attribute similarity.With factor analysis method,the weight of each item as well as the comprehensive score can be obtained.Then,the network matrix is established by calculating similarity between users.According to the method of cliques division,the subjects are divided into fractions.And eventually,friend recommendation service is completed.[Result/conclusion]Through the empirical analysis,it concludes that the relationship-attribute recommendation,with implement of cliques division,does better,especially,in the condition where the length of recommendation lists is limited,and efficiently enhances precision.
出处
《情报理论与实践》
CSSCI
北大核心
2020年第2期137-142,163,共7页
Information Studies:Theory & Application
基金
2017年广东省哲学社会科学基金项目“基于Altmetrics的学术成果多维信息计量体系、评价模型及实证研究”(项目编号:GD17CTS01)
2018年国家社会科学基金项目“基于用户行为动机的ALTMETRICS评价模型构建与实证研究”(项目编号:18BTQ075)的系列成果
关键词
好友推荐
用户属性
关系相似度
派系划分
社交网络
因子分析
friend recommendation
users’characteristics
relationship similarity
cliques division
social network
factor analysis
作者简介
余以胜,男,1975年生,博士,副教授,硕士生导师。研究方向:电子商务与信息经济。;陈咏晖,女,1997年生,硕士生。研究方向:电子商务,消费者行为研究。
