Research And Application Of Task Assignment Method In Spatial Crowdsourcing

Crowdsourcing is a distributed computing paradigm that integrates the collective intelligence of the society to accomplish complex tasks that computers or nonprofessionals cannot complete.The formation of the sharing economy society has proposed the emergence and development of a location-based crowdsourcing service application——spatial crowdsourcing.It has contributed a great deal of value in life areas such as convenient life and intelligent transportation.The basic and core research directions in spatial crowdsourcing include task assignment and user privacy protection.However,a large number of previous studies only focused on crowdsourcing tasks and workers,and ignored the multi-skill task assignment that requires knowledge of different specialized fields.Secondly,as spatial crowdsourcing is a location-based service,it needs to rely on the actual location of task executor to complete task assignment.The untrusted platform server and malicious network attackers pose potential threats to users' location privacy information.Based on the shortcomings of the above research work,the main research contents of this thesis are shown as follows:Firstly,an assignment problem of complex spatial tasks based on multi-skills is proposed in the single-task scenario,and we defined an optimization problem of maximizing the task utility by considering such factors as skill constraints,task requirements on the number of fine-grained workers and worker capability limitation.Considering that there is no effective optimal solution method for this problem,three approximate solution algorithms based on greedy strategy are proposed.The validity and correctness of the algorithms are verified by experiments on real data sets and synthetic data sets.Secondly,on the basis of the above research in the single-task scenario,the assignment problem of complex spatial tasks based on multi-skills in multi-task scenario is further proposed.Because in real life,more than one requester publishes the task,and workers with a wide range of expertise or interests may be needed for more than one task.In order to solve this conflict,the optimization goal is redefined to maximize the total utility of the system.By analyzing the influence of different factors on the experimental results,the feasibility and correctness of the scheme are verified.Finally,to solve the problem of user location privacy disclosure in the process of spatial crowdsourcing task assignment,we proposed a privacy protection model satisfying differential privacy.The model does not require the intervention of any trusted third party organization and can provide personalized privacy protection level for workers.The same as the above research content,in order to better adapt to the practical spatial crowdsourcing application,we introduce the skill constraints of workers and tasks.In the multi-task and multi-worker scenario,the problem of skill matching security task assignment based on differential privacy technology is proposed.By proving that the problem is NP hard,a task assignment algorithm based on greedy strategy is designed,and it is proved that the scheme provides the security of differential privacy protection.At last,the effectiveness of the proposed algorithm is verified by comparative experiments on real data sets.