Multi-Campaign Oriented Spatial Crowdsourcing

Libin Zheng, Lei Chen

Abstract: Recently, spatial crowdsourcing has been drawing increasing attention with its great potential in collecting geographical knowledge. The system throughput (number of assigned tasks) and workers' travel distance are two of many important factors in spatial crowdsourcing, and the improvement to one of them usually means the sacrifice of the other. However, most existing works resolve the trade-off between these two factors by simply targeting tasks within a bounding circle of each worker. In this paper, we compromise between the throughput and the distance by formulating these two factors as score terms in the objective function. This flexible formulation has the advantages of abandoning distant tasks and minimizing workers' travel distance for reachable tasks. Aside from that, we study the multi-campaign scenario of spatial crowdsourcing, which is not uncommon in practical applications while not yet discussed in existing works. The worker diversity of the campaigns is considered to be another goal and formulated as another score term in the objective function. Subsequently, the problem of multi-campaign oriented spatial crowdsourcing is to maximize the objective function comprised by the aforementioned score terms. We prove that the problem is NP-hard, thus, we propose several approximation solutions. Extensive experiments have been conducted to confirm the effectiveness and the efficiency of the devised solutions.