考虑声誉效应的网络众包参与者行为博弈模型及仿真分析

为了控制网络众包交易的风险,以网络众包平台中的发包方和接包方为研究对象,基于微分博弈方法建立了网络众包交易中参与者行为的博弈模型,研究了发包方和接包方在合作和非合作两种情形下的策略选择。本文研究发现在声誉效应满足特定阈值的条件时,合作博弈情形下众包系统的总收益更大;而且,发包方和接包方违约控制的收益及努力程度更大。在网络众包平台的违约罚金超过特定阈值时,随时间推移发包方和接包方违约控制的努力程度逐渐提高,实现帕累托改进。本文结合猪八戒网、InnoCentive等网络众包平台的众包交易机制,采用数值分析对博弈模型的求解结果进行模拟。对于网络众包平台来说,可以通过建立有效的信用评价机制,将信用评价或声誉纳入众包交易定价,引导参与者约束自身行为。

Participants' Behavior Game Model and Simulation Analysis in Crowdsourcing Considering Reputation Effect

In order to prevent the risk in crowdsourcing, this paper focuses on participants, including employers and service providers in crowdsourcing, and constructs non-cooperative and cooperation game model of the participants' behavior in crowdsourcing based on differential game theory. Then we explore different game strategies of employers and service providers under different scenarios. The result shows that when reputation effect gets threshold condition, the return of employers and service providers and the total return of crowdsourcing system in the case of cooperative game are greater than those of non-cooperative game. Similarly, the effort controlling default behavior of the employers and service providers in the case of cooperative game are is also greater than that of non-cooperative game. When the penalty imposed by platform gets a certain threshold, the effort controlling default behavior of employers and service providers increases gradually to achieve Pareto improvement. The crowdsourcing platform can incorporate credit rating or reputation into the crowdsourcing transaction pricing by establishing an effective credit assessment mechanism to guide participants in restricting their behavior and controlling risks in crowdsourcing transaction. In addition, we set the model parameter based on Zhubajie crowdsourcing platform as case, and make a numerical analysis to validate the theoretical model.