Fair Task Assignment in Spatial Crowdsourcing

With the pervasiveness of mobile devices, wireless broadband and sharing economy, spatial crowdsourcing is becoming part of our daily life. Existing studies on spatial crowdsourcing usually focus on enhancing the platform interests and customer experiences. In this work, however, we study the fair assignment of tasks to workers in spatial crowdsourcing. That is, we aim to assign tasks, considered as a resource in short supply, to individual spatial workers in a fair manner. In this paper, we first formally define an online bi-objective matching problem, namely the Fair and Effective Task Assignment (FETA) problem, with its special cases/variants of it to capture most typical spatial crowdsourcing scenarios. We propose corresponding solutions for each variant of FETA. Particularly, we show that the dynamic sequential variant, which is a generalization of an existing fairness scheduling problem, can be solved with an O(n) fairness cost bound (n is the total number of workers), and give an O( n m ) fairness cost bound for the m-sized general batch case (m is the minimum batch size). Finally, we evaluate the effectiveness and efficiency of our algorithm on both synthetic and real data sets.