An Intelligent Job Recommendation System based on Semantic Embeddings and Machine Learning

Abstract

To address the shortcomings in existing approaches of job recommendation systems, this paper proposes a novel machine-learning-based job recommendation system that performs bi-directional matching for dynamic and accurate recommendations. The proposed approach generates ideal job recommendations for a targeted Curriculum Vitae (CV) and vice versa. Unlike previous approaches, the proposed approach incorporates natural language processing (NLP) techniques to extract linguistic features such as Bag of Words (BoW), n-grams, TF-IDF, and Parts-of-Speech (PoS) tag and build a rich feature set. These features are further analyzed using semantic embeddings, enabling robust job matching. Experiments were performed to validate the performance of the proposed approach. The designed system is validated on various real-world datasets, overcoming the dataset size limitations of prior works. Due to combination of semantic embeddings, machine learning, and various similarity measures, this approach demonstrates the potential to deliver reliable, explainable, and ideal job recommendations, addressing the challenges of static and false outputs in existing systems.