Research Area:  Machine Learning

Abstract:

With the growth of online recruitment, hiring platforms receive and store an enormous amount of resumes and job posts of various fields. Hence, applying automatic resume classification in database systems can reduce the time and labor required for database management, allowing recruiters to re-update the databases quickly. In past studies on resume classification and machine learning techniques, the shortage of labeled resume datasets used in model training was a problem to be addressed. Therefore, this study applied a domain adaptation approach based on a graph neural network, so that the latent features of job posts can be extracted to classify resumes without requiring labeled resumes and a retraining model. The domain adaptation approach compares the semantic similarity between job posts and resume contents for resume classification. The proposed method was tested in an actual recruitment database and achieved a highly accurate result.

Keywords:  

Author(s) Name:  Thi-Thuy-Quynh Trinh,Yu-Chi Chung, R.J.Kuo

Publisher name:  Elsevier BV

DOI:  10.1016/j.knosys.2023.110364

Volume Information:  volume 266,2023

Paper Link:   https://www.sciencedirect.com/science/article/abs/pii/S0950705123001144