| Summary: | Online social network has become a major medium for information propagation. Users are also confused by huge amount of information read every day. It is important to find a solution to judge the authenticity of a certain piece of information and figure out the rumor source, which is also our focus. In this paper, we transfer the social network data into an adjacent matrix, while each element of the matrix symbolizes the directed edge between two nodes. We described the process of rumor source detection by selecting a small group of nodes as monitors. Their work is to monitor data flow. We have classified the monitors into two kinds: one has heard the information called positive monitors, and the other one hasn’t heard information called negative monitors. After the ranking of four metrics about rumor source distance with monitors, we can figure out rumor source ID. The accuracy of the result depends on the number of monitors we select and the methods we use to select monitors.
For rumor assessment, it is a reasonable situation that an information generated by a small group of clustered spreaders tend to be rumor, while an information like news will have a wide range of independent source. So we adopt a greedy source set algorithm to figure out the original source set. Information started by smaller source set has a high possibility to be a rumor.
|
|---|