基于活跃度的众包工作者信誉模型

针对现有众包系统不能有效地控制众包交互过程中工作者的活跃积极性和任务完成质量的问题，提出了一种基于活跃度的工作者信誉模型来实现众包平台的质量控制。该模型改进了平均信誉模型，从工作者活跃度和历史信誉值的角度提出了活跃因子和历史因子的概念。首先根据众包工作者最近30 d内参与众包活动的天数计算工作者的活跃因子；然后根据历史因子计算众包工作者的历史信誉值；最后根据计算出来的活跃因子和历史信誉值计算基于活跃度的工作者信誉值，以衡量众包工作者的工作能力。理论分析和测试实验结果表明：与平均信誉模型相比，根据基于活跃度的工作者信誉模型选取的众包工作者在任务完成质量上提高了4.95%，在任务完成时间上减少了25.33%；与基于证据理论信任模型相比，在任务完成质量上提高了6.63%，在任务完成时间上减少了25.11%。实验结果表明，基于活跃度的工作者信誉模型在实际众包项目中能够有效提高众包任务的完成质量，减少众包任务的完成时间。

Reputation model of crowdsourcing workers based on active degree

Aiming at the problem that the existing crowd-sourcing system can not effectively control the active enthusiasm of the workers and the quality of task completion in the process of crowd-sourcing interaction, a worker reputation model based on active degree was proposed to realize the quality control of the crowd-sourcing platforms. The model improved the average reputation model, and the concepts of active factor and historical factor were put forward from the point of view of workers' active degree and historical reputation value. First, the active factor of the worker was calculated according to his participating days in the crowd in the last 30 days, and then the historical reputation value of the crowd-sourcing worker was calculated according to the historical factor. Finally, the reputation value of the crowd-sourcing worker based on active degree was calculated based on the calculated active factor and historical reputation value, which was used to measure the ability of the crowdsourcing worker. The theoretical analysis and test results showed that compared with the average reputation model, the task completion quality of crowdsourcing workers selected by the worker reputation model based on active degree was increased by 4.95% and the completion time was decreased by 25.33%; compared with the trust model based on evidence theory, the task completion quality was increased by 6.63% and the completion time was decreased by 25.11%. The experimental results show that the worker reputation model based on active degree can effectively improve the quality of crowdsourcing tasks and reduce the completion time.