Two-sided Online Stable Task Assignment with Incomplete Lists and Ties in Spatial Crowdsourcing

Weiyi Huang, Peng Li, Bo Li, Jing Liu, Lei Nie, Haizhou Bao

Abstract: For the application of task assignment in crowdsourcing, previous research has focused on offline version or one-sided online version. Some of them draw extra attention to stability in task assignment. Our work extends a one-sided task assignment and its stable generalization. In particular, we present a two-sided online stable task assignment (TS-OSTA) problem in which any vertex from two sides can maintain an incomplete neighbor list and ties. We first give the fundamental definitions of task assignment in spatial crowdsourcing and stable matching. Then, we analyze and present the matching condition and formulate the TS-OSTA problem. As a warm-up, we investigate and analyze two classical baseline algorithms Greedy and Random Threshold. Then we present a bipartite ranking algorithm that maintains the efficiency of Greedy and effectiveness of Random Threshold. Through extensive experiments, we examine the performance of our algorithm in size, memory, and time. The results show that our algorithm performs better than the baselines.