Um sistema para ranqueamento e previsão de similaridade entre vagas de emprego e candidatos

Abstract

This study explores the use of machine learning in recommendation systems applied to recruitment, highlighting the potential of Graph Attention Networks (GATs) to model relationships between candidates, job openings, and companies. A GAT-based model was developed and evaluated on a real dataset, focusing on ranking and similarity prediction tasks. The results demonstrated superior performance in global metrics and significant computational efficiency, with execution times reduced by up to fourfold in the largest dataset subset. However, limitations such as the absence of temporal information and the use of random negative sampling indicate opportunities for improvement. This research contributes to the advancement of re commendation systems in the recruitment context, providing a foundation for future studies that incorporate temporal data and more refined training methods.