Online Social Network Information Dissemination Integrating Overconfidence and Evolutionary Game Theory

Abstract

Public opinion inversion and other nonlinear phenomena often occur during online social network information dissemination (OSNID) in modern society. To explore the influence of overconfidence on OSNID, we develop a multi-agent simulation model integrating overconfidence and evolutionary games, as previous research has scarcely paid any attention to irrational behavior in information dissemination. This integrated paradigm provides an effective tool for managers to control the spread of OSNID and reduce the harm caused by rumors. The theoretical model is constructed from the perspective of an evolutionary game and, combined with overconfidence theory, three overconfidence scenarios are designed: benefit overconfidence, cost overconfidence, and both benefit and cost overconfidence. Then, a multi-agent simulation model of OSNID under different overconfidence scenarios is realized. The proposed simulation model exhibited better performance (e.g., faster diffusion speed) than the traditional Bass model; its performance was also validated by comparison with real-world cases. The results demonstrate that (1) with increasing benefit overconfidence, the convergence speed of OSNID will be accelerated, and the user group will reach stability at a faster speed. (2) With increasing cost overconfidence, the user’s decision will change from adopting dominant to tit-for-tat, and the adoption ratio will gradually decrease. (3) Compared with overconfidence in earnings, overconfidence in costs is more conducive to improving stability. This study attempts to provide new ideas for OSNID and attempts to propose an integration framework for behavioral theory and simulation methods.

Keywords

• Evolutionary game
• information dissemination
• multi-agent simulation
• online social network
• overconfidence theory