Machine Learning Techniques for Matching Candidates’ Profiles with Job Advertisements

Aleksandra Davidova Stefanova, Tosin Adewumi

Keywords: Job-Resume Matching, Machine-Learning, Embeddings, Sentence Transformer, Cosine Similarity, TF-IDF

TL;DR: A comparison between machine learning models and traditional keyword matching algorithms in resume-job matching tasks using a public dataset from Kaggle and a private corporate dataset

Abstract: This study presents a comparison between machine learning models and traditional keyword matching algorithms in resume-job matching tasks using a public dataset from Kaggle and a private corporate dataset. The results showed that the Ada and T5-based models generated the highest cosine similarity scores, which align with human judgment. This suggests that traditional keyword matching algorithms are not as well-suited for the task as much as state-of-the-art (SotA) deep learning models.

Serve As Reviewer: ~Saadane_Rachid1, ~Paul_D'Alessandris1, ~H.Steven_Scholte1, ~J.R.Corney1

Submission Number: 20