Answers are grounded in this paper and cite their sources.
System evaluation demonstrates automated layout-aware parsing and hybrid semantic ranking across diverse resume formats, indicating improved efficiency for high-volume recruitment.
This paper presents an AI-driven resume-ranking web application tailored to enhance large-scale job recruitment processes by leveraging cutting-edge deep learning and natural language processing (NLP) techniques. The system integrates multiple advanced models, including YOLOv9 for layout-aware segment detection, EasyOCR for high-accuracy text recognition, GLiNER for named entity recognition, and a fine-tuned multilingual BERT (mBERT) model for text classification. These components operate in tandem to extract and semantically interpret candidate information from resumes submitted in various formats such as PDF and DOCX. A custom dataset was developed specifically for training the object detection model, where YOLOv9 achieved a superior mean Average Precision (mAP) of 0.84. For semantic matching, the system adopts a hybrid approach combining dense vector similarity via gte-large-en-v1.5 embeddings and cosine similarity, alongside BM25-based keyword matching for relevance scoring. This dual strategy ensures accurate and contextually aware resume-to-job matching. The user-friendly web application allows HR professionals to upload resumes, define job criteria, and receive ranked results, thereby significantly reducing manual effort. Despite challenges in text extraction from irregular layouts and zero-shot NER limitations, the system exhibits robust performance, showcasing the promise of AI-powered recruitment tools in automating and improving candidate selection.
No takes yet. Share an insight, caveat, or question.
TEJA et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: