Inference of Rumor Sources in Networks

We study the problem of inferring rumor sources in networks. We start with a simple rumor spreading model and then construct estimators for the rumor source and provide theoretical bounds for their performance. In the process we develop a new notion of network centrality which we term rumor centrality, which is an exact maximum likelihood estimator for regular trees. Experimental results from synthetic and real world networks verify the good performance of our rumor source estimators.

IT School Poster 2009.pdf — PDF document, 4721Kb