在线社交网络中有向社交超图数据放置策略  

Directed Social Hypergraph Data Allocation Strategy in Online Social Networks

作　　者：杨文茵[1,2] 王国军[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]佛山科学技术学院电子与信息工程学院,广东佛山528000

出　　处：《小型微型计算机系统》2015年第7期1559-1564,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61472451,61272151)资助;教育部博士点基金项目(20110162110043)资助

摘　　要：减少访问请求开销、改善副本效益和负载均衡是提高在线社交网络数据放置性能的重要目标,当前的相关研究多侧重对某个指标进行优化,并且所使用的二元社交图未能直观有效地反映社交网络多用户交互的特性.综合考虑三个指标的相关因素,提出一个有向社交超图模型,并基于该模型提出一个两阶段社交感知数据放置方案,其中包括社交超图均衡划分算法和社交性流行度感知的副本复制放置算法.使用Facebook数据集进行的实验表明,该策略比同类方案综合性能指标更优,实现了上述三个目标的平衡.Reduction of the query cost,improvement of replication efficiency and load balancing are significant goals of performance improvement in data allocation for Online Social Networks（ OSNs）. However,existing models and methodologies for OSNs data allocation seldom comprehensively consider all these goals,and the common dyadic social graph could not properly model the features of user interactions in OSNs. We propose a directed social hypergraph model,on which we performs a two-phase data allocation algorithm. Data partitioning and replication would take fully into account of factors including social locality,data popularity,server workload and network environment,etc. The experimental evaluation of Facebook data set shows that the scheme we put forward achieves a good tradeoff among objectives of lowquery cost,high replication efficiency and load balancing.

关 键 词：在线社交网络 数据放置 有向社交超图 社交热度感知 数据复制 负载均衡 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]