An Online Task Assignment Based on Quality Constraint for Spatio-Temporal Crowdsourcing

Abstract

Crowdsourcing is the perfect embodiment of group wisdom. With the rapid development of mobile network and the sharing economy model, spatio-temporal crowdsourcing technology has been research hotspot. Task assignment is one of the core issues of spatio-temporal crowdsourcing technology. There are three algorithms: Random Algorithm, Random-Threshold-Based Algorithm (RT) and Adaptive random-threshold-based Algorithm (Adaptive RT) for maximizing the total utility in the online task assignment of three types of objects, tasks, workers and workplaces. But these algorithms ignore the distance cost and fairness between task requester and workers. Unfairness means that higher task's reward with lower worker's success ratio or lower task's reward with higher worker's success ratio in a match. Therefore, this paper proposes Quality Constraint Algorithm (QCA), which quantifies fairness between task requester and workers as match quality and adopts a matching strategy of automatic negotiation on task's reward to improve the average match quality. QCA not only has higher average match quality and higher total utility, but also optimizes the average distance cost. Compared with Adaptive RT, QCA has an average increment of 11% on total utility, an average increment of 19% on average match quality and an average decrease of 17% on distance cost. In term of time cost, QCA is only 8% of Adaptive RT.