Efficient Worker Recruitment Approach Based on Social Incentive Diffusion in Mobile Crowdsourcing

Haojun Teng, Zhongwei Zhan, Yingjie Wang, Wenhan Hou, Yiheng Pan, Zhipeng Cai

Keywords: Mobile Crowedsourcing, Worker Recruitment, Social Network, Optimization Algorithm, Graph Neural Network

Abstract:

With the advancement of mobile devices and wireless communication technology, Mobile Crowdsourcing (MCS) has emerged as a paradigm for recruiting workers to complete tasks at specific locations.
As a variant, social MCS has been proposed to facilitate large-scale collaborative tasks among multiple workers, leveraging MCS social network to expand the worker pool and enhance task utility. This paper introduces the Worker Recruitment approach based on Social Diffusion (WRSD). To model and incentivize workers' social diffusion behavior towards task information, we designed a Social Trust Prediction (SoTP) framework using Graph Convolutional Networks (GCNs) and a social incentive mechanism. For effective worker-task matching, we integrated cross-modal social recommendation data using Graph Attention Networks (GATs) within a Social Task Recommendation (SoTR) framework. The WRSD problem is then modeled as a Constrained Multi-attribute Combinatorial Optimization (CMCO) problem under budget constraints.
We define a heuristic neighborhood search strategy and propose the Variable Neighborhood Tabu Search (VNTS) algorithm to solve the WRSD problem, achieving an approximately optimal worker solution for each task. Comprehensive experiments conducted on three real-world datasets validate the effectiveness and efficiency of the proposed approach.

Submission Number: 77