Stable Task Assignment for Socially Aware Collective Computing Systems
Hao Zhou, Yang Li, Yunlong Zhao, Chunyan Liu · 2024
Collective Computing is the latest generation of computing paradigm that has received a lot of attention recently. One of the important things in Collective Computing systems is how to assign tasks to human participants. When doing the assignment, the platform may face the challenge of inadequate availability of skilled workers. To deal with this challenge, some researchers leverage the property that workers are socially aware and consider assigning tasks to workers with the assistance of social networks. However, the objectives of task requesters and workers can conflict and may result in unstable assignments due to unhappy requesters and workers, which is ignored by most of these works. This paper leverages the influence propagation on the social network to assist task assignment and takes requesters’ preferences and workers’ preferences into consideration. Stable Matching Theory is leveraged to solve the problem of matching tasks with seed workers and an efficient task assignment algorithm is proposed. Through extensive simulations, the performance of the proposed algorithm is evaluated in different settings. Empirical studies show that the proposed algorithm outperforms the baseline algorithms in achieving stability and system efficiency.