Nextgenresume: Enhancing Resume Classification with Large Language Models

Automated resume classification has become increasingly important as recruitment processes increasingly rely on intensive resume analysis. However, resumes comes in doc and pdf format, where pdf format is used often, requiring a reliable PDF extraction method to standardize data for processing. Rule-based extraction and keyword matching algorithms difficult to generalize across resume formats, hence they sometimes produce inferior results. The proposed approach uses Large Language Models (LLM) for its ability to understand and process natural language, addresses the challenge of obtaining organized résumé information and connecting it with job descriptions. The biggest challenges are managing multiple formats, acquiring experience data such dates and firm names, and guaranteeing Resume-Job Description (JD) relevance. The proposed approach use transformer-based embeddings for Resume-JD alignment, customized Named Entity Recognition (NER) for structured data extraction, and pdfplumber for PDF extraction. Similarity evaluates resumes for relevance, whereas LLM model DistilBERT contextualizes resumes and job descriptions. Challenges include addressing a typical resume forms, improving the NER model for entity recognition, and optimizing embeddings for computational performance. The proposed process involves preparing job descriptions and resumes, extracting relevant sections using NER and regular expressions, tokenizing the language, embedding using DistilBERT, and scoring matches using cosine similarity. The proposed resume-job description matching accuracy of 92.4% was far better than before and automates candidate shortlisting, saving time in recruitment.