Selection Methods And Application Research For Mobile Crowdsourcing Based On Ability Evaluation

Mobile devices have become more and more popular in human's lives in recent years,which provide abundant resources and development prospect for the application of mobile crowdsourcing(MC).MC is a burgeoning method for resource collection and sharing.it can collect resources and information from dispersive individual mobile device and accomplish many tasks that can hardly be achieved by the traditional ways.For the MC system,the key for providing good MC services is to select proper workers for the task.The present research exist the following challenges:1)there are few researches about how to build an evaluation model between the MC task and MC worker,thus it can't properly evaluate worker's ability to execute task and whether the result provided by worker can conform to the requirement of task requester.2)Researches of MC selection for static scenario are usually divided to platform-only way and worker-only way,the former may bring some risk of privacy disclosure,and the latter will bring heavy overhead to the computation and communication.While in the dynamic MC scenario,the task needs to wait for a while to recruit enough workers.It's a problem demanding prompt solution for how to promote the quality of MC services by selecting workers for task in the dynamic scenario.3)In the MC application,there are few researches about building a reasonable MC selection mechanism to provide MC services,it's a problem to study for how to use MC selection methods to improve the service quality of MC application.In allusion to the aforementioned challenges in MC researches,this thesis conducts relevant researches on the MC ability evaluation,MC selection methods and the MC application based on the ability evaluation.Specifically speaking,the main work of this thesis includes the following aspects1)To make a quantitative assessment for the correlation between MC task and MC worker,this thesis firstly propose an ability evaluation model for the MC system This model defines the system model of MC,and it calculates the ability value for the worker performing task by using worker's familiar degree to the keywords set and the keywords set's importance degree to the task,meanwhile we define the ability adjustment scheme and the computing method for the hidden information.2)For the static MC selection scenario,this thesis proposes a static MC selection method based on worker ability assessment model.This method makes two MC selections both on the platform and the mobile device of worker,which ensures finding workers with higher ability value while using workers' sectional information,and it can guarantees a lower computation and communication overhead.For the dynamic MC selection scenario,this thesis proposes a dynamic MC selection method based on worker ability assessment model.This method takes the time properties for worker and task,and it selects workers based on threshold value and priority.We conduct experiments for these two methods,and the results demonstrate the effectiveness of our methods3)To apply the evaluation model and selection methods we proposed into practice,this thesis designs an MC application for electric demand forecasting in allusion to the scenario that uses MC in the forecasting field.The application uses the ability evaluation model and selection methods proposed in this thesis,and it can provide service for the MC activities such as data cleaning,information collection and load forecasting.And we introduce the using process by a specific example.