Positive Information Diffusion for Rumor Containment in Online Social Networks

Rumors in online social networks (OSNs) create social chaos, financial losses, and endanger property, which makes rumor containment an important issue. We consider an OSN in which the users communicate via private peer-to-peer messages. We propose a peer-to-peer linear threshold (PLT) model for information diffusion in OSNs, which is a variant of the classic LT model. To combat the rumor spread in the OSN, an anti-rumor agent introduces positive information to a few users of the OSN, which are called as positive seed nodes. These positive seeds spread the positive information in the OSN in due course of time. For a given time-period called the rumor-relevance interval, we determine average number of rumor-influenced nodes for the random, the max-degree, the greedy, and the proposed proximity-weight-degree based positive seed node selection schemes. We compare the effect of the rumor-relevance interval duration and number of positive seed nodes on the average number of rumor-influenced nodes for different positive seed selection algorithms. Our experimental results show that proximity-weight-degree based positive seed selection algorithm performs on par with the high complexity greedy scheme.