GKD-Recruiter: Jointly Modeling Social and Task Heterogeneity for Spatial Crowdsourcing via Graph Knowledge Distillation

Yucen Gao, Zhemeng Yu, Zhuoran Li, Jianxiong Guo, Xiaofeng Gao

Abstract: Social recruitment offers a solution to worker scarcity in Spatial Crowdsourcing (SC) but faces challenges that are often ignored in traditional Influence Maximization. First, task heterogeneity arising from offline execution constraints breaks the interest-implies-participation assumption, as social influence often fails to translate into physical presence. Second, finite task demand creates a saturation trap, a non-submodular setting in which utility drops sharply to zero once demand is met. To bridge these gaps, we propose GKD-Recruiter, a Task-Aware framework designed to maximize Effective Task Satisfaction (ETS). We explicitly model the complex worker-task affinity via a heterogeneous graph and capture directional social influence using a novel Influential GAT. To robustly fuse these distinct signals, we introduce a Graph Knowledge Distillation mechanism. Furthermore, we employ Rainbow DQN to navigate the non-submodular combinatorial search space, avoiding the local optima that trap greedy heuristics. Extensive experiments on real-world datasets demonstrate that GKD-Recruiter significantly outperforms state-of-the-art baselines in both solution quality and inference efficiency. The code is available at https://github.com/GaoYucen/GKD-Recruiter.

Lay Summary:

Location-based applications like food delivery, ride-sharing, and localized crowdsourcing often struggle to find enough workers in certain areas. To solve this, platforms frequently use social referral bonuses, encouraging current workers to invite their friends. However, traditional "viral marketing" methods designed for online information spreading fail in this scenario. First, a socially influential person might successfully get their friends to read a post, but they cannot easily convince them to physically travel to a specific location to complete a task. Second, each physical job only needs a limited number of people; over-inviting workers wastes the platform's budget.

In this paper, we introduce a smart AI model called GKD-Recruiter to overcome these challenges. Instead of just looking at who is popular in the social network, our model learns to match people's social influence with their actual willingness and physical ability to complete specific real-world tasks. By using an advanced learning approach, it also plans ahead to avoid wasting invitations on jobs that are already filled. Tested on real-world data, our method helps platforms find the right workers much faster and more cost-effectively, ultimately improving the efficiency and availability of real-world crowdsourcing services.

Link To Code: https://github.com/GaoYucen/GKD-Recruiter

Primary Area: Social Aspects

Keywords: Crowdsourcing, Worker Recruitment, Influence Maximization

Submission Number: 14463