Refining IEFP´s recommender systems: a comparative exploration of project based learning and a modernized framework

Abstract

This study explores an alternative approach to skill-based recommender systems to match job candidates and opportunities, avoiding the usage of dimensionality reduction methods on skills. Using several evaluation metrics, it compares the results of this new approach with results obtained by the previous models. The proposed approach of splitting the skills data into one dataset per major group of a job proves to be more efficient when comparing to the previous methods. Despite modest evaluation results, the model shows promise. Overall, this research concludes that such a recommendation system has great potential to enhance matching diversity within Public Employment Centers.