An evolutionary game model for the spread of non-cooperative behavior in online social networks

Abstract

We present an evolutionary game-theoretic model for the spread of non-cooperative behavior in online social networks. We formulate a two-strategy game wherein each player's behavior is classified as normal (cooperate) or abusive (defect) and pairwise interactions between adjacent players in the network graph yield a unique payoff to each according a prisoner's dilemma payoff structure. Player strategies evolve by imitation of successful behavior in observable neighborhoods. We demonstrate convergence of player behavior over time to a final strategy vector. Proof-of-concept is given for a real-world dataset collected from a popular online forum.