基于网络博弈动力学的反恶意众包研究

Crowdsourcing can break through various kinds of limitations and barriers with lower cost and higher efficiency, and thus promote the optimal allocation and sharing of the innovative resources and elements. The openness of the internet can make every individual the outsourcee, which also provides the chances for adversaries to easily sabotage the crowdsourcing outcome with low cost and low risk. Obviously, the malicious behavior in crowdsourcing, known as “Malicious Crowdsourcing”, severely hinders the large-scale applications of crowdsourcing. However, it is challenging to thwart the malicious crowdsourcing due to the power-law differences of individuals’ behaviors, the multidimensional heterogeneous associations among the individuals and the asymmetry of the information. Hence, based on the network game dynamics, in this project, we will propose a series of frameworks and mechanisms mainly focusing on: 1) the anti-malicious crowdsourcing mechanism based on the zero-determinant strategy; 2) the outsourcer-dominant mechanism design game for anti-malicious crowdsourcing; 3) the outsourcee-assisted mechanism design game for anti-malicious crowdsourcing. Our study uses the original problems of pricing and task allocation in crowdsourcing as the weapons against malicious behaviors, by which the solutions are embedded in the process of implementing crowdsourcing. Thus the solution and implementation are integrated organically and systematically to filter the wicked with an economic method, realizing the Pareto optimality and finally promoting the healthy development of crowdsourcing.

众包能以较低成本、较高效率跨越各类局限与壁垒，推动创新资源和创新要素的优化配置与共享。与此同时，互联网的开放性使得任何别有用心的人都可能成为接包方，以低成本低风险的恶意行为毁掉众包成果。这种众包恶意行为被称为“恶意众包”，它极大地阻碍了众包的大规模应用。然而，个体行为的幂律差异、个体之间的多维异构关联、信息的不对称性，使得反恶意众包充满挑战。为此，本项目拟基于网络博弈动力学理论，提出一套框架和机制，重点研究：1）基于零行列式策略的反恶意众包机制； 2）发包方主导的契约型反恶意众包机制；3）接包方参与的契约型反恶意众包机制。上述研究利用众包本需解决的定价和任务分配问题作为反恶意众包的武器，将解决手段“内嵌”到众包的执行过程，使两者形成有机、系统的整体，以经济手段除恶扬善，实现全局帕累托最优，促进众包的健康发展。

众包能以较低成本、较高效率跨越各类局限与壁垒，推动创新资源和创新要素的优化配置与共享。但互联网的开放性使得任何别有用心的人都可能成为接包方，以低成本低风险的恶意行为毁掉众包成果。这种众包恶意行为被称为“恶意众包”，它极大地阻碍了众包的大规模应用。然而，个体行为的幂律差异、个体之间的多维异构关联、信息的不对称性，使得反恶意众包充满挑战。为此，本项目拟基于网络博弈动力学理论，提出一套框架和机制，重点研究：1）基于零行列式策略的反恶意众包机制，揭示恶意众包的多维成因，量化分析个体和集群恶意行为的行动规律和演化过程，抑制接包方恶意行为同时最大化双方收益；2）发包方主导的契约型反恶意众包机制，所设计的契约满足个体特征，具有评估精准性和泛用性，可实现非合作博弈模型中发包方针对专家评价系统的反恶意众包最优控制； 3）接包方参与的契约型反恶意众包机制，针对性的设计满足激励相容、参与约束、反共谋且尊重接包方意愿的契约，防止众包中的谎报和共谋，实现无验证的质量控制和资源的最佳配置。