A Cross-View Hierarchical Graph Learning Hypernetwork for Skill Demand-Supply Joint Prediction
Wenshuo Chao, Zhaopeng Qiu, Likang Wu, Zhuoning Guo, Zhi Zheng, Hengshu Zhu, Hao Liu
TL;DR
The paper tackles the volatile alignment of job market skills by proposing CHGH, a cross-view hierarchical graph learning Hypernetwork that jointly predicts skill demand and supply. It combines a cross-view graph encoder to capture inter-view asymmetric relationships, a hierarchical graph encoder to model cluster-level co-evolution, and a conditional hyper-decoder that leverages historical demand-supply gaps to jointly predict future shares. The model is trained with a composite objective including a clustering loss and an L2 regularization term, and discretizes outputs into five trend classes for robust classification performance. Across three real-world datasets, CHGH consistently outperforms seven baselines, with ablation studies confirming the value of each module and case analyses aligning predictions with observed market shifts. The approach offers a principled, data-driven pathway for individuals and organizations to anticipate evolving skill requirements and address potential gaps in the labor market.
Abstract
The rapidly changing landscape of technology and industries leads to dynamic skill requirements, making it crucial for employees and employers to anticipate such shifts to maintain a competitive edge in the labor market. Existing efforts in this area either rely on domain-expert knowledge or regarding skill evolution as a simplified time series forecasting problem. However, both approaches overlook the sophisticated relationships among different skills and the inner-connection between skill demand and supply variations. In this paper, we propose a Cross-view Hierarchical Graph learning Hypernetwork (CHGH) framework for joint skill demand-supply prediction. Specifically, CHGH is an encoder-decoder network consisting of i) a cross-view graph encoder to capture the interconnection between skill demand and supply, ii) a hierarchical graph encoder to model the co-evolution of skills from a cluster-wise perspective, and iii) a conditional hyper-decoder to jointly predict demand and supply variations by incorporating historical demand-supply gaps. Extensive experiments on three real-world datasets demonstrate the superiority of the proposed framework compared to seven baselines and the effectiveness of the three modules.