An Efficient Bi-Encoder-Based Skill Classification and Employer Notification Model for Smart Job Market Analysis

The modern labour market requires efficient analysis of unstructured job advertisements to identify required skills and job roles. However, many existing labour market analysis systems rely on large language models (LLMs), which are computationally expensive and difficult to deploy in real-time environments. To address this limitation, this project proposes an AI-enabled Job Portal System that integrates lightweight natural language processing techniques for intelligent recruitment analysis. The system utilizes a bi-encoder–based contrastive learning approach inspired by ConTeXT-match to extract relevant skills from job advertisements and map them to standardized taxonomies such as ESCO. In addition, JobBERT V2–based job title normalization ensures consistent classification of diverse job titles. The platform allows applicants to register their educational and skill profiles, which are automatically matched with job requirements. A Gmail-based notification system provides real-time updates regarding application status. The proposed system improves recruitment transparency, computational efficiency, and scalability for smart labour market analysis.