Dynamic Knowledge Graphs: Revolutionizing Skill Analytics through Graph Neural Networks

Abstract

The dynamic nature of workforce skills and their interrelationships necessitates sophisticated systems for understanding, classifying, and predicting skill evolution. This article introduces a novel framework for Dynamic Knowledge Graph Evolution in the Skills Domain, leveraging Graph Neural Networks to model hierarchical skill relationships, analyze temporal trends, and automate taxonomy generation. Our architecture incorporates a hierarchical GNN model that captures parent-child relationships among skills, enabling accurate skill classification and clustering. Temporal article is integrated into the framework to identify emerging skills and trends, providing actionable insights for dynamic workforce planning. Furthermore, we propose an automated taxonomy generation algorithm that leverages unsupervised clustering and natural language processing to continuously update the graph structure. Experiments on real-world datasets demonstrate the framework's ability to model skill evolution with high precision, generate interpretable taxonomies, and predict workforce demands effectively. Applications in talent management, career planning, and learning systems showcase the system's ability to adapt to the rapidly evolving landscape of workforce skills.