论文部分内容阅读
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.
Information dissemination and detection is given a communication network, how to choose the most effective set of nodes as the observation node or deployment of sensors to detect as soon as possible the spread of the network information, which for the social network leader in opinion mining, rumors spread detection, public opinion Monitoring and other applications is of great importance.In this paper, we propose a ranking algorithm based on random walk model to solve DiffRank, which is based on the characteristics of network structure, node content and historical propagation data. According to the result of this algorithm, -k nodes as observing nodes to detect the possible spread of information in the network.Experimental results based on the real data of Sina Weibo show that, compared with other similar algorithms, the DiffRank algorithm in the detection of coverage, detection time and the number of people infected with information decreased 3 indicators are better than similar algorithms.As for the scalability of the algorithm, DiffRank algorithm is more suitable for parallel or distributed computing, scalability is better.