Task Assignment And Resource Optimization For Crowdsourcing

With the development of Internet,mobile Internet,Internet of Things and other related technologies,the amount of data in various industries is growing rapidly.People urgently need an effective means to fully mine the value behind the massive data and have a more comprehensive and in-depth understanding of the data.The emergence of crowd intelligence and crowdsourcing provides more possibilities to solve this problem.In order to manage tasks more effectively,many crowdsourcing platforms have emerged at home and abroad,such as Amazon's Mechanical Turk and Net Ease's Youdao crowdsourcing.Through these platforms,users can easily publish and accept a variety of tasks.However,due to the uncertainty of workers' behavior,the platform can not directly deliver tasks to workers randomly.Therefore,it is necessary to build a reasonable quantitative model of tasks and workers to achieve accurate task assignment,which has become a hot and difficult issue in the academic field.At the same time,more and more users participate in the application of spatial crowdsourcing.In order to reduce maintenance cost of massive user data and improve Service Level Agreements,various crowdsourcing service providers use cloud to store data effectively.However,considering the largescale,wide area distributed workers and massive unstructured data in the actual spatial crowdsourcing application scenario,it is difficult to provide economic and reliable data storage services only by using single cloud,and there are risks and challenges such as vendor lock-in and privacy leakage.Therefore,in the multi cloud environment,how to combine the characteristics and requirements of spatial crowdsourcing to provide an effective data placement scheme is an urgent challenge.Therefore,based on the above issues,this paper conducts the following research on how to make the reasonable tasks assignment for crowdsourcing platforms and how to manage spatial crowdsourcing data cost-effectively in a multi-cloud environment:(1)Suitability-based task assignment in crowdsourcing markets.This paper uses the picture classification task as the starting point to study the active assignment of knowledge acquisition tasks in the crowdsourcing market.In this paper,the process of crowdsourcing task assignment is abstracted into a weighted bipartite graph matching model,and determines the difficulty of tasks by introducing the Word Net external knowledge base.At the same time,it refers to the rank mechanism of e-sports and uses the dynamic update strategy to evaluate the ability of workers in real time.On this basis,the weighted Euclidean distance with penalty factor is used to calculate the suitability,which is used to measure the degree of fitness between tasks and workers.Finally,Kuhn-Munkres(KM)algorithm is used to solve the model,so as to assign tasks with different difficulties to workers with matching abilities and improve the overall quality of the tasks.(2)Cost-effective and latency-optimal data placement strategy for spatial crowdsourcing in multi-cloud environment.This paper studies the problem of finding effective placement solutions for spatially distributed crowdsourced data in a multi-cloud environment to minimize costs and access latency.The paper fully considers the interval pricing strategy adopted by different CSPs,analyzes the geographical distribution characteristics of data centers with the help of DBSCAN density clustering algorithm,and proposes an effective data initialization placement strategy.Finally,the improved genetic algorithm is used to further optimize the results.(3)Design and implementation of crowdsourcing service solution platform.Based on the task assignment for crowdsourcing platform and the optimization of spatial crowdsourcing data storage in the multi-cloud environment,this paper constructs a web-based one-stop crowdsourcing service solution platform from the perspective of crowdsourcing service providers.The platform mainly includes functional modules such as crowdsourcing task assignment and spatial crowdsourcing data storage.The implementation of the platform provides theoretical support and technical verification for crowdsourcing service providers to carry out reasonable task assignment and cost-effective user data management,which has certain practical value.