Research On The Astroturfing Behavior Based On Evolutionary Game

Nowadays in e-commerce websites,astroturfing's behavior is more variable and hidden,which brings greater challenges to the supervision of astroturfing.The current research on astroturfing's behavior at home and abroad is mainly divided into two categories: analysis based on the characteristics of astroturfing's behavior and analysis of interactive behavior based on game theory.The former is still a blacklist method in essence,which causes the technical supervision always lagged behind the discovery of astroturfing's behavior,it's difficult to grasp the initiative of supervision;the latter is mainly based on player's entire rationality,which ignores the fact that in fact,each player is bounded rationality,with incomplete information and insufficient computing power,it is impossible to capture the dynamic interaction process by only one game.Therefore,we focus on the incomplete information characteristics of multiple roles,and research on astroturfing's behavior based on evolutionary game.The main innovations and contributions of this paper are as follows:(1)Aiming at the current research of astroturfing's behavior based on game theory does not fully analyze the complete process of decision-making evolution of players,this paper constructs online astroturfing supervision game model between "online platforms-sellers" based on evolutionary game theory.we analyze the interest demands and game relations between online platform and seller in the online astroturfing supervision game,builds on the assumptions of parameters based on the actual situation,construct the online astroturfing supervision game model based on evolutionary game theory,then we use replicator dynamic to analyze the decision-making evolution process of platform and seller,finally obtain the evolutionary stability strategy of the system.And through analysis,we obtain the key factors that affect the astroturfing supervision.Finally,simulation experiments are designed to verify the validity of the analysis results and conclusions,so as to provide a theoretical basis for the supervision of astroturfing in reality.(2)In long-term astroturfing's supervision,it is difficult for platform to balance supervision cost and supervision effect.We design eight supervision strategies based on "Iterated Prisoner's Dilemma",which are: Tit for Tat(TFT),Contrite Tit for Tat(CTFT),Never Forgive(NF),Tit for Two Tat(TF2T),Tit for Three Tat(TF3T),Soft Majority(SM),Hard Majority(HM)and Win-Stay Lose-Shift(WSLS).Through simulation experiment between platform and seller,we obtain the dynamic regulatory strategy adjustment mechanism of platform under different probabilities of not hiring astroturfing and different detection accuracy.Through calculation,the platform's income when adopting the dynamic regulatory strategy adjustment mechanism is compared with when adopting the fixed regulatory strategy,the minimum increase is 5.9%,the maximum increase is 303%,and the average increase is 130.2%,which proves the effectiveness of the dynamic regulatory strategy adjustment mechanism.