On inferring rumor source for SIS model under multiple observations

• Published in: International Conference on Digital Signal Processing proceedings pp. 755 - 759
• Main Authors: Wang, Zhaoxu, Zhang, Wenyi, Tan, Chee Wei
• Format: Conference Proceeding Journal Article
• Language: English
• Published: IEEE 09.09.2015
• Subjects: Algorithms; Approximation algorithms; Computational modeling; Conferences; Detectors; Diffusion; Digital signal processing; Heuristic algorithms; Information dissemination; Joints; maximum likelihood detection; Network topology; Networks; Online social networks; rumor source detection; Silicon; SIS model; statistical inference; Trees; Unions
• ISSN: 1546-1874; 2165-3577
• DOI: 10.1109/ICDSP.2015.7251977

This paper studies the problem of a single rumor source detection based on the susceptible-infected-susceptible (SIS) spreading model. Based on the rumor centrality proposed in the Susceptible-Infected (SI) model by Shah and Zaman, we propose a rumor centrality based algorithm, that leverages multiple observations to first construct a diffusion tree graph, and then use the union rumor centrality to find the rumor source. Our simulation results on different network structures shows that our proposed algorithm performs well. For tree networks, increasing the observations can dramatically improve the exact detection probability. This clearly indicates that a richer diversity enhances detect-ability.