Balancing quality and budget considerations in mobile crowdsourcing

Mobile/spatial crowdsourcing is a class of crowdsourcing applications in which workers travel to specific locations in order to perform tasks. As workers may possess different levels of competence, a major research challenge for spatial crowdsourcing is to control the quality of the results obtained. Although existing mobile crowdsourcing systems are able to track a wide range of performance related data for the participating workers, there still lacks an automated mechanism to help requesters make key task allocation decisions including: 1) to whom should a task to allocated; 2) how much to pay for the result provided by each worker; and 3) when to stop looking for additional workers for a task. In this paper, we propose a budget-aware task allocation approach for spatial crowdsourcing (Budget-TASC) to help requesters make these three decisions jointly. It considers the workers' reputation and proximity to the task locations to maximize the expected quality of the results while staying within a limited budget. Furthermore, it supports payments to workers based on how their track records. Extensive experimental evaluations based on a large-scale real-world dataset demonstrate that Budget-TASC outperforms the state-of-the-art significantly in terms of reduction in the average error rate and savings on the budget.