A multi-criteria decision making based integrated approach for rumor prevention in social networks

Rumors are unverified pieces of information whose veracity status is unknown at the time of circulation. Preventing them at an early stage helps to mitigate the consequential loss and reduces the cost of recovering the nodes affected by rumors in the social network. However, existing research focuses more on rumor detection and control than on rumor prevention. In this article, we propose a Multi-Criteria Decision Making (MCDM) integrated approach for rumor prevention that addresses three issues. Firstly, the need for a rumor prevention model to debunk them at an early stage. Secondly, introduction of counter-rumor in the system as soon as a rumor is detected to reduce the time lag between rumor diffusion and counter-rumor diffusion. Thirdly, selection of trusted authorities as key nodes to initiate the counter-rumor diffusion. The proposed method is two-fold. Firstly, we propose Analytic Hierarchy Process - Technique for Order of Preference by Similarity to Ideal Solution (AHP-TOPSIS) method, an MCDM-based approach, to select the important nodes and host agents on them. Secondly, using agent nodes, we propose a counter-rumor diffusion model Susceptible-Infected-Recovered-Prevented-Agent (SIRPA) which is an improved variant of the popular epidemic model Susceptible-Infected-Recovered (SIR). The proposed SIRPA model is simulated on six social network datasets and multiple experiments are performed on these networks to validate the proposed work. The results show that our proposed SIRPA model is effective in preventing rumors in social networks and outperforms the baseline model.