摘要
针对社交网络环境中,为用户推荐哪类好友会使用户更容易采纳问题,提出一种社交网络中基于角色活跃度的好友推荐方法,该方法结合了社交网络环境中不同社群(团队)拓扑结构形成的社群角色同社群中同样角色不同用户行为形成的角色活跃度差异和用户兴趣做好友推荐.首先通过文本相似性为用户寻找兴趣的社群,然后利用E-GARGO模型构建了社群拓扑结构中角色活跃的定义,并给出了活跃度计算方法,根据计算方法为目标用户推荐活跃度在Top-N的好友推荐列表.在构建的推荐机制上,以学者为交互背景的科研社交网站"学者网"为应用背景,通过网页问卷调查的方式得出好友推荐平均准确率比原网站有所改善,并且以5分制的意见采纳度做考察中得出了平均值为4.3030和4.5152的采纳度.
To maximize the probability that a user would accept the recommended candidates, the paper proposes an friend recommen- dation mechanism based on users＇ interest, users＇ activity in different sub-community and the topology of sub-network. Firstly, through text similarity, get interested sub-communities. Secondly, with the help of the E-GARGO model, we formulate the definition and com- puting method of roles＇ activities. Based on this, the values of roles＇ activities are computed in the users＇ interested sub-communities. Then the Top-N friends among the candidates whose value exceeds the target user are recommended. Through a questionnaire survey, the proposed friend recommendation mechanism has an improvement in the recommended accuracy and the average accuracy, and ac- celerate the probability of acceptance.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第10期2162-2165,共4页
Journal of Chinese Computer Systems
基金
国家"八六三"高技术研究发展计划项目(2013AA01A212)资助
国家自然科学基金项目(61272067
61502180)资助
广东省公益研究与能力建设专项(S2012030006242)资助
广东省自然科学基金项目(2015A030310509
2014A0303010238)资助
