Rumor Source Identification in Social Networks with Time-varying Topology (Feb-2018)
The main aim of this project is to identify the real rumor source and eliminate fake message and rumor source in the social Network.
In this proposed method we propose a novel source identification method to overcome the challenges. First, to represent a time-varying social network, we reduce it to a sequence of static networks, each aggregating all edges and nodes present in a time-integrating window. This is the case, for instance, of rumors spreading in Bluetooth networks, for which the fine-grained temporal resolution is not available, whose spreading can be studied through different integrating windows _t (e.g., _t could be minutes, hours, days or even months). In each integrating window, if users did not activate the Bluetooth on their devices (i.e.,offline), they would not receive or spread the rumors. If they moved out the Bluetooth coverage of their communities (i.e., physical mobility), they would not receive or spread the rumors. Second, similar to the detective routine in criminology, a small set of suspects will be identified by adopting a reverse dissemination process to narrow down the scale of the source. The reverse dissemination process distributes copies of rumors reversely from the users whose states have been determined based on various observations upon the networks. The ones who can simultaneously receive all copies of rumors from the infected users are supposed to be the suspects of the real sources.