Research And Implementation Of Task Undertaking Algorithm And Task Allocation Algorithm On Decentralized Crowdsourcing Platform

With the rapid development of the Internet and knowledge economy,crowdsourcing has achieved rapid development due to its open resource aggregation and low cost.However,on the one hand,the existing crowdsourcing platforms are all centralized crowdsourcing platforms,whose centralized architecture mechanism makes it difficult to guarantee the fairness,privacy and reliability of crowdsourcing transactions.On the other hand,task allocation algorithm is also an important research field of crowdsourcing,which is related to whether tasks can be completed with high quality on time.In view of the above problems,the main research results of this paper are as follows:1)Taking data annotation crowdsourcing as an example,proposes a decentralized crowdsourcing platform application based on Edgence,which is an existing decentralized application deployment framework,and designs its architecture and mechanism in detail.In this decentralized crowdsourcing platform,a masternode is a kind of block chain node deployed on the edge server,and its upper layer is built with decentralized crowdsourcing application.The masternode plays the role of decentralized intermediary in crowdsourcing transactions,reducing the dependence of crowdsourcing transactions on a single intermediary,and improving the reliability,privacy and fairness of the crowdsourcing platform.2)The new application scenario,task undertaking and allocation process of the mastemodes on the decentralized crowdsourcing platform is modeled as an online learning problem.Firstly,Thompson sampling algorithm is introduced to construct the characteristics of users.Secondly,a task undertaking algorithm is proposed to maximize expected revenue.The expected revenue is estimated based on the user's schedule,credit value and the matching degree between the user's field of expertise and the task field of expertise,and then the task undertaking algorithm based on maximize expected revenue is made.After that,a task allocation algorithm based on maximize revenue is proposed.With the set of received tasks and the set of subordinate users as the vertex,and the estimated task revenue as the edge weight,a weighted bipartite graph is established.Then,Kuhn-Munkres algorithm is introduced to solve the optimal matching problem.Finally,through the simulation experiment,the performance superiority of the proposed algorithm in task completion rate,task completion quality and average revenue of masternode is proved.