DEVELOPMENT OF A MODULE FOR INFORMATION EXTRACTION AND CURRICULUM VITAE CLASSIFICATION FOR CANDIDATE MATCHING USING MACHINE LEARNING MODELS

• Main Author: Risqi Firdaus, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/85034
• Institution: Institut Teknologi Bandung

The job application process is complex for both applicants and companies, especially when the number of applicants for a single job posting can reach thousands. Companies need significant resources to screen documents and conduct interviews. Therefore, an intelligent system is required to optimize the job application process, making it more efficient. One of the breakthroughs to address this issue is the development of an Intelligent System for Candidate Matching and IT Talent Assessment. This system is designed to handle the entire recruitment process, from registration and document screening to assessment or interviews. This thesis aims to develop one of the modules within this Intelligent System, specifically the Information Extraction and Job Application Document Classification Module using Machine Learning Models. The module is designed to assist human resources teams in screening job application documents. The information extraction module is developed using a machine learning approach. The transformer-based model, trained with transfer learning techniques, achieved the best performance with a recall score of 0.81. Meanwhile, the document classification model for matching job applications with company requirements showed the best performance using the SVM model, with an AUC score of 0.85 and a precision score of 0.70. Although the model is considered reliable based on its AUC score, its performance, particularly in terms of recall, still requires improvement.