Influence Evaluation And Incentive Mechanisms For Crowdsoursing Task Diffusion Based On Social Network

Crowdsourcing is an open and distributed problem-solving method.The mobile crowdsourcing platform sends out task invitations to users using mobile devices via the Internet to complete tasks that are challenging for machines.Crowdsourcing has become an efficient problem-solving model in the rapidly developing Internet society.The flexibility of crowdsourcing provides unprecedented opportunities to a wide range of applications.However,the lack of participants limits the development of crowdsourcing.In order to solve this problem,this thesis proposes a global influence evaluation and its incentive method for crowdsourcing task diffusion based on social networks.It mainly includes the following two aspects:In order to solve the problem of insufficient participants of the crowdsourcing,this thesis proposes a global influence evaluation method based on the crowdsourcing task diffusion on social networks,so as to make full use of the propagation of personal influence in social networks to recruit more people for crowdsourcing.At present,there are many researches on social influence considering that each user in a social network is only affected by their social neighbors.But in real life,the users,especially key figures in various fields,have their influence on the global network.Therefore,this thesis estimates the global influence based on the shell decomposition and uses it in the task diffusion model.The simulations based on real-world data sets show that the social cost of the task diffusion algorithm using the proposed evaluation method is much lower than the Fast-Selector algorithm,but the task completion rate is also lower than the Fast-Selector algorithm.Aiming at the diffusion of crowdsourcing tasks,existing work ignores the characteristics of diffusion tasks.This thesis designs a fine-grained crowdsourcing task diffusion system to select the registered users according to their influence on different task topics.First,the crowdsourcing task diffusion system is modeled as a reverse auction,and the budget feasible task diffusion problem is formalized.From the perspective of achieving higher matching degree between the tasks in the crowdsourcing system and the task diffusers,this thesis proposes a topic-aware independent cascading task diffusion model,which closely connects the influence of the task diffusers with the task topics.A parameter estimation algorithm based on maximum likelihood estimation is proposed to estimate the parameters in the above model.Furthermore,for the problem of budget-feasible task diffusion,a computationally effective,individual rational,budget-feasible and truthful incentive mechanism is proposed.Through theoretical analysis and simulation experiments,this thesis shows that the proposed budget feasible mechanism obtains more platform value and effective user tasks than the comparision algorithms,and obtain a better task diffusion effectiveness.