DiffRank:一种新型社会网络信息传播检测算法

DiffRank:A Novel Algorithm for Information Diffusion Detection in Social Networks

信息传播检测是给定一个传播网络,如何选择最有效的节点集合作为观察节点或部署传感器,以尽早尽快检测到网络中传播的信息,这对于社会网络中的意见领袖挖掘、谣言传播检测、舆情监控等应用具有重要意义.文中结合网络结构特点、节点内容属性、历史传播数据等信息,提出了一个基于随机游走模型的传播能力排序算法DiffRank,根据该算法的结果选择传播能力最强的top-k个节点作为观察节点来检测网络中可能出现的信息传播.基于新浪微博真实数据的实验结果表明,与其他同类算法相比,DiffRank算法在检测覆盖率、检测时间和信息感染人数下降比率3个指标上,都优于同类算法.在算法的可扩展性方面,DiffRank算法更加适用于并行或分布式计算,可扩展性更好.

Given a social network,information diffusion detection can be modeled as selecting aset of nodes as observations to detect the spreading of information or rumors as quickly as possible.Itcan be well applied to fields like opinion leader detection,rumor detection,and public security.Incorporating network structure,node attribute and history information cascades,we propose arandom walk based algorithm DiffRank to sort nodes according to their diffusion ability,thenchoose the top-犽nodes on the list as observations to detect information diffusion.Experiments onreal dataset of Sina Weibo show that DiffRank outperforms other algorithms with respect toinformation cascades coverage ratio,detection time and reduction of infected population.Besides,DiffRank can be implemented easily in distributed or parallel computing environment,achievinggood scalability.