Balancing Competition for Fairness-Aware Task Recommendation and Assignment in Spatial Crowdsourcing

With the rapid evolution of sensing techniques, spatial crowdsourcing (SC) has gained increasing attention in both academia and industry. SC aims to assign location-based tasks to mobile workers, where task recommendation assists workers in identifying appropriate and appealing tasks. However, existing studies mainly focus on worker satisfaction, diversity, or coverage in task recommendation, overlooking the competition among workers, i.e., multiple workers simultaneously contend for the same task after recommendation. Excessive competition may result in low assignment success rates, while insufficient competition undermines overall profitability. To address these problems, we study a novel problem of Competition-balanced Task Recommendation for Fair Assignment (CTRFA) in SC, where tasks are first recommended to multiple workers, workers then select tasks based on their individual willingness for task assignment. CTRFA aims to maximize profit and the assignment success rate by balancing competition among workers, ensuring fairness. We develop an innovative Supply-Demand Adjustment Task Recommendation-to-Assignment framework, which encompasses two major components: task recommendation and task assignment. In the task recommendation, we propose a novel Multistage Probabilistic Recommendation Algorithm to maximize the profit and design a Supply-Demand Flow Balancing Algorithm to maximize the assignment success rate, which constructs a Supply-Demand Transfer Graph to identify the most appropriate task regions for each worker balancing competition among workers. Further, we introduce a Fairness-aware Assignment Algorithm, which prioritizes task assignment to workers with lower assignment success rates enabling fairness. Experiments on two real datasets offer insight into the effectiveness of the proposals, showing up to $\mathbf{1 1. 6 \%}$ performance improvement.