ATS Resume NLP Analyzer: A Hybrid, Explainable, and Practical Framework for Resume-Job Matching

Abstract: This paper proposes and experimentally validates a production-ready hybrid framework for automated resume–job matching using explainable Natural Language Processing (NLP) techniques. The system processes heterogeneous resume formats and job descriptions to compute an interpretable Applicant Tracking System (ATS) score by integrating deterministic skill matching, semantic similarity using sentence embeddings, and lexical keyword overlap.

We construct a modular pipeline that transforms unstructured resume text into structured representations, enabling robust comparison against job requirements. The system employs a hybrid scoring mechanism combining three signals: skill coverage, embedding-based semantic similarity, and token-level keyword matching. Additionally, the framework includes OCR-based ingestion for scanned resumes, explainable score decomposition, missing-skill diagnostics, and optional Large Language Model (LLM)-assisted feedback generation.

Experimental evaluation is conducted on a curated dataset of resumes and job descriptions with human -labeled relevance scores. The proposed system achieves strong alignment with human judgment, demonstrating improved ranking consistency over baseline keyword-only approaches. The system maintains low latency suitable for real-time deployment and includes robust fallback mechanisms for production reliability.

This work demonstrates that a hybrid deterministic-semantic approach can significantly improve transparency, usability, and effectiveness in automated recruitment systems while remaining scalable and deployable in real-world environments.

Keywords: Applicant Tracking System, Resume Screening, NLP, Semantic Similarity, Explainable AI, Recruitment Analytics, FastAPI