Multi-Model Semantic Skill Gap Analysis System for AI Resume Evaluation

This paper presents a hybrid semantic and multi-model approach for intelligent skill gap analysis in workforce informatics, with a focus on AI-driven resume evaluation. Traditional recruitment systems primarily rely on keyword-based matching, which fails to capture contextual relationships between skills and often leads to inaccurate candidate-job matching. To address these limitations, the proposed system integrates transformer-based models such as BERT and Sentence-BERT (SBERT) with multi-modal data sources, including resumes, job descriptions, and project repositories. The methodology involves a structured pipeline consisting of data acquisition, preprocessing, feature extraction using Natural Language Processing (NLP) techniques, semantic embedding generation, multi-modal fusion, and skill gap detection using cosine similarity. The system computes a skill gap score to evaluate candidate suitability and provides personalized recommendations to help users improve their skill sets in alignment with industry requirements.