Transforming Tech Hiring: BERT-Powered Neural Networks for Automated Job Title Classification

In the development of the technology era, the ever-expanding job market exposes an intimidating challenge for job seekers to navigate through the opportunities of securing their desired job positions. Hence, it is of utmost importance to precisely identify suitable job titles for job descriptions, skills and qualifications relating to individual candidates during the recruitment process. The proposed methodology aims to classify the job titles within the domain of IT-focused job dataset utilizing Word Piece tokenization following the Bidirectional Encoder Representations from Transformers (BERT) embedding with Convolutional Neural Network (CNN) for multi classification. The comparative study demonstrates that the proposed approach, combining BERT embedding with CNN, achieves approximately 86% accuracy and F1 score outperforming traditional classification models, neural networks, and BERT-based classification models.