Incentivize Mechanisms On Budget Constraint Crowdsourcing Systems

Nowadays,crowdsourcing systems and crowdsensing models are subtly affecting people's daily life and work.In 2006,the editor of Wired magazine Jeff Howe1 proposed the concept of crowdsourcing for the first time.Since then,more and more researchers have started to explore and study incentive mechanism in the crowdsourcing systems.Today,there is a well-established theoretical system for offline research on the incentive mechanism in crowdsourcing systems,while fewer researches on the design of online incentive mechanisms.Incentive mechanism is reflected in the process of task distribution,the platform side can not fully grasp the ability of all employees,only the current time has reached the user's ability to be known by the platform side.This paper proposes an online incentive mechanism framework for crowdsourcing systems.Different from the traditional offline incentive mechanisms,it requires that the assignment of tasks should be done without fully grasping all the information of the users,so that the platform can allocate as much as possible many tasks under the condition of budget balance,at the same time to ensure that the tasks assigned users meet certain qualities,in order to maximize the benefits of the platform.In order to solve this problem,this paper presents a method based on the combination of knapsack and secretary problem.The algorithm is mainly divided into two phases:user observation phase and user selection phase.In user observation phase,the incentive mechanism does not directly reject all users but select them based on their declared performance.In user selection phase,the algorithm will select the threshold according to the previous arrived users and the threshold will be updated dynamically after the success of winner selection.To better analyze the performance of this mechanism,this paper presents the feasibility of analyzing and evaluating the algorithm with the game-theoretical properties in auction theory,such as: individual rationality,truthfulness,computational efficiency etc.Based on these properties,this paper has performed theoretical analysis.Then,with the proposed online incentive mechanism,this paper builds a simulation platform based on Matlab,analyzes the performance of mechanism using synthetic data,and compares with other common algorithms of classical incentive mechanisms.Simulation results show that the incentive mechanism has great improvement under different criteria.At the same time,it also validates that the proposed algorithm is in line with the above game-theoretical properties.