Automated Difficulty and Algorithmic Tag Prediction for Competitive Programming Problems Using Natural Language Processing

This research proposes a lightweight NLP-based framework for automatically predicting the difficulty rating and algorithmic tags of competitive programming problems. The system combines TF-IDF text features with Random Forest Regression for difficulty prediction and One-vs-Rest Logistic Regression for multi-label tag prediction. The models are evaluated on unseen Codeforces problems, showing that traditional NLP and machine learning techniques can provide useful support for problem analysis, practice selection, and educational recommendation systems.