Research On Competing Crowdsourcing Task Assignment Mechanisms With Preference

As a novel task distribution and execution paradigm based on Internet,crowdsourcing has become an effective tool that utilizes human intelligence to perform tasks that are challenging for machines.The incentive mechanism is important for the crowdsourcing system to stimulate workers,and to achieve good service quality.In a crowdsourcing system,there may be the competition among the task requesters,and the requesters may show the preference for the workers.However,few existing incentive mechanisms have ever considered the issue of stimulating the competing and biased requesters for the crowdsourcing system.In this thesis,we explore truthful task assignment mechanisms to maximize the total value of accomplished tasks for this novel scenario.First,we present three models of crowdsourcing: Identical workload Identical value Model(II-Model),Identical workload Non-identical value Model(INModel),Non-identical workload Non-identical value Model(NN-Model).These system models take the competition among the requesters,the preferences of the requesters and the workload constraint of the workers into consideration.Our first model assumes that the workload and value for each task are identical.The second model is a generalization of the first one,considering non-identical value of tasks.Our third model considered the non-identica workload and non-identica value of the tasks.Furthermore,we design an assignment mechanism for each of the three models: TAM-II(Truthful Assignment Mechanism for the II-Model),TAM-IN(Truthful Assignment Mechanism for the INModel)and TAM-NN(Truthful Assignment Mechanism for the NN-Model),each assignment mechanism follows the matching approach to solve the Valuation Maximizing Assignment(VMA)problem.Through both rigorous theoretical analyses and extensive simulations,we demonstrate that the proposed assignment mechanisms can achieve computational efficiency,workload feasibility,preference truthfulness,and constant approximation.