A Stable Task Assignment Scheme in Crowdsourcing
Xiao Chen · 2019
Crowdsourcing has become a popular business development strategy that outsources self-contained small tasks to a crowd of people to solve problems that an individual or an organization cannot easily do. There are two different task assignment types in crowdsourcing platforms: the worker selected task mode and the server assigned task mode. Right now, the crowdsourcing websites only use one of them and the satisfaction of the workers and the requesters is not fully addressed. Furthermore, it is not easy for the requesters to identify qualified workers quickly. In this paper, we propose a crowdsourcing model that considers the preferences of both the requesters and workers to improve their satisfaction and thereafter benefits the crowdsourcing platform. We first put forward a ranking formula for the requesters to identify qualified workers timely based on the Bayesian inference by considering two factors: the prices the workers charge and their online reviews, and then propose a stable task assignment algorithm STA that stably matches the workers and the tasks through the stable marriage approach. Simulation results show that our proposed task assignment approach greatly improves the satisfaction of the requesters and the workers compared with the existing Hungarian method and the STA variations that only consider one factor.