Research On Task Assignment With Guaranteed Privacy And Budget For Crowdsourcing Platforms

Crowdsourcing combines the collective intelligence of a large number of workers to carry out tasks which are difficult to perform by individuals.For a specific task,recruiting the optimal subset of workers is the key to successfully accomplish the task.In other word,worker recruitment(task assignment)is a most important issue for crowdsourcing platform.Tasks have different quality requirements and budget constraints.The task is accomplished successfully when the aggregate quality is not less than the requirement,otherwise fails.Workers participate in the task and are paid by the crowdsourcer,and the payment depends on the skill level of workers.Both workers and tasks have different preferences,task prefers high-level workers,and workers prefer cost-effective tasks.How to achieve a stable matching between tasks and workers is facing great challenges.At the same time,workers involved in crowdsourcing tasks appeal to protect their online privacy.For example,location information implies workers’ quality of life,home address,identity,etc.,such that attackers may cause adverse effects on workers after stealing such information.Therefore,the dual guarantee of stability of matching result and user online privacy is the primary goal of crowdsourcing task assignment.In this thesis,taking the stability of allocation results and the privacy protection of users into account,we conducts a comprehensive study on task assignment.The main contributions are introduced as follows:(1)Since the traditional crowdsourcing task assignment framework does not protect workers’ privacy,we leverage the budget constraints of tasks,the requirements of quality,the skill level of workers as well as the location relationship between tasks and workers,and then propose a crowdsourcing task allocation algorithm,TA-auction,based on online auction and smart contract.Online auction can ensure the stability of task assignment results,and smart contract can protect the privacy of workers.Before task assignment,the auctioneer decides the amount of deposit according to the information submitted by the workers.The workers pay the deposit to the smart contract.The workers who do not deposit can not participate in the crowdsourcing task assignment.Reasonably,the smart contract confiscates the deposit of the workers who quit midway or do not implement the assignment result.During task assignment,the crowdsourcer recruits the workers who are geographically close to the specific task,and adopts the second price auction to decide the winner,and achieve the stable matching between workers and tasks.Experimental results and performance analysis show that TA-auction scheme can protect users’ privacy and guarantee the stability of allocation results.(2)Focus on the execution time constraints of tasks,we explore the task execution time requirements and the available execution time of workers,and propose an task allocation algorithm with multiple constraints,online TA-auction.Online task allocation is realized by smart contract,which can not only protect users’ privacy,but also prevent users from reneging.Firstly,a grouping algorithm is designed according to the location relationship between tasks and workers.Secondly,a task allocation algorithm is designed based on the second price auction,and a stable allocation result is obtained finally.Experimental results and performance analysis show that online TA-auction can achieve better social welfare and higher worker satisfaction with low computing cost and low communication cost.