Optimal Containment of Misinformation in Social Media: A Scenario-Based Approach

Abstract

The rapid expanding of online social networks (OSNs) in terms of sizes and user engagements have fundamentally changed the way people communicate and interact nowadays. While OSNs are very beneficial in general, the spread of misinformation or rumors in OSNs not only causes panic in general public but also leads to serious economic and political consequences. Several studies have proposed strategies to limit the spread of misinformation via modifying the topology of the diffusion networks, however, a common limit is that parameters in these diffusion models are difficult, if not impossible, to be extracted from real-world traces. In this paper, we focus on the problem of selecting optimal subset of links whose removal minimizes the spread of misinformation and rumors, relying only on actual cascades that happened in the network. We formulate the link removal problem as a mixed integer programming problem and provide efficient mathematical programming approaches to find exact optimal solutions.