Intelligent Resume Parser and Classifier Using Regular Expressions

This paper presents a Resume Parser and Classifier that effectively extracts key details and classifies resumes by integrating compiler design concepts like regular expressions with Machine learning(ML) models. The system has two phases: Lexical Analysis and Classification. Lexical analysis is done using regular expressions to extract key details like name, contact information, skills, and education. By converting these details into feature vectors using TF-IDF and One-hot encoding, two datasets are created. By converting the extracted text from resume images into Word2Vec embeddings, a third dataset is created. A comparative analysis is done between datasets and models by giving the datasets as input to five ML models: RF, SVM, KNN, XGB and LGBM. These models predict the most suitable job role from 14 classes. To handle class imbalance, SMOTE is applied. PCA is employed for reducing dimensionality in order to find the optimal number of components. The system is evaluated using accuracy and f1-score as performance metrics. RF has proven to be the best model with an accuracy of 98.29% and Fl-Score of 98.30%. The combination of traditional compiler design principles for preparing the datasets with ML provides an effective tool for resume parsing and classification, offering a reliable and feasible solution to aid recruiters by accurately matching candidates to job roles.