Modeling And Research On The Formation Of Crowdsourcing Teams Considering Uncertainty Of Task

In recent years,crowdsourcing that is an emerging group collaboration model has received much attention in various fields,and the group intelligence that it adopts can significantly improve the efficiency of task completion.In crowdsourcing scenarios such as software development and copywriting,the characteristics of uncertainty are widespread,including the following three types.The first type is the uncertainty of task requirements.Task requirements(category,workload)are often dynamically adjusted by requesters during the execution of tasks.The second type is the uncertainty of the arrival time associated with the tasks.The time that a task appears in crowdsourcing systems is unpredictable.The last type is the uncertainty of revenue gained by completing the task.Requesters may pay money depending on the quality of task completion that is associated with the quality of the work,which brings the uncertainty of the revenue of the task completion.Previous researches usually assume that the requirements and revenues of tasks are deterministic.Although some existing researches have focused on the uncertainty of the arrived time of tasks,it usually ignores the structures of formed teams when dealing with new tasks,which results in the higher costs of team formation.To complete tasks with uncertainty,we systematically analyze the impacts caused by the three kinds of uncertainty on team formation,and propose team formation algorithms to improve the efficiency of task completion.An incentive mechanism of team formation that satisfies the constraints of cost is designed for the case where there is uncertainty in the requirements of tasks.The mechanism selects workers to satisfy the requirements of tasks based on marginal contribution,and develops an incentive strategy to ensure the honesty of workers.The experiments show that our method can reduce the cost of task completion in this scenario.A team collaboration mechanism is designed to minimize the cost of negotiation,where the teams collaborate with each other to complete tasks.We also develop a strategy to ensure individual honesty.The proposed mechanism and strategy ensure the efficiency of task completion and minimize the cost of team formation.Experiments show that our method can reduce the cost of completing task and improve the efficiency in this scenario.Algorithms are proposed to maximize social welfare in the context that there is uncertainty in revenue gained by completing tasks.First,for the case that each task has the same mode of payment,we propose an allocation algorithm based on dynamic programming,and prove the optimal substructure and transition equation.Second,aimed at the general case,we design a heuristic method based on greedy idea.The experiments show that the proposed methods can improve social welfare in these scenarios.