TWO STAGE JOB TITLE IDENTIFICATION SYSTEM FOR ONLINE JOB ADVERTISEMENTS

Abstract

The rapid growth of online recruitment platforms has led to an enormous increase in unstructured job advertisement data, creating challenges in accurately identifying and standardizing job titles due to variations, inconsistencies, and noisy textual information. This project proposes a Two Stage Job Title Identification System for Online Job Advertisements to improve the extraction, classification, and normalization of job titlesfrom large-scale recruitment data. In the first stage, job advertisements collected from various sources such as online job portals, company websites, and recruitment databases undergo preprocessing using Natural Language Processing (NLP) techniques, including tokenization, stop-word removal, part-of-speech tagging, and named entity recognition, to identify potential job title candidates. In the second stage, machine learning and deep learning models such as Support Vector Machine (SVM), Random Forest, and transformer-based architectures like BERT are employed to classify and refine the extracted job title candidates into standardized occupational categories. The system further maps synonymous and variant job titles, such as “Software Developer” and “Software Engineer,” into a unified taxonomy, reducing ambiguity and improving consistency. By transforming unstructured recruitment data into structured information, the proposed framework enhances job search accuracy, i mproves employer– candidate matching, and supports labor market analytics. Experimental results demonstrate that the two-stage approach significantly outperforms conventional single-stage methods in terms of precision, recall, and overall classification accuracy. The system provides a scalable, intelligent, and efficient solution for job title standardization and information retrieval in modern online recruitment environments.