Volume 2 - Issue 2, March - April 2026

Previous Year Question (PYQ) papers serve as valuable resources for students preparing for examinations. However, manually analyzing multiple question papers to identify frequently repeated questions and important topics is timeconsuming and error-prone. This paper presents PrepVision AI, an intelligent question paper analysis system that leverages Natural Language Processing (NLP) techniques and semantic similarity algorithms to automate this process. The proposed system accepts multiple question papers in PDF or image format, extracts questions using Optical Character Recognition (OCR) and PDF parsing, performs comprehensive text preprocessing including instruction filtering, and applies TF-IDF vectorization with cosine similarity to group semantically similar questions. The system identifies repeated questions across papers, ranks them by importance using a multi-factor scoring algorithm considering frequency, topic relevance, keyword density, and question complexity, and generates two outputs: An Important Questions Report categorizing questions by importance levels, and a Practice Question Paper organized into examination sections. Experimental evaluation demonstrates effective instruction noise removal, accurate semantic question grouping at 65% similarity threshold, and meaningful importance ranking. PrepVision AI provides students with data-driven insights for focused exam preparation while maintaining transparency by presenting analysis-based practice papers rather than guaranteed predictions.

Rohail Rafiq Shaikh, Savli Sumit Patil, Vedant Santosh Sargar, Komal Ashok Jagtap, "PrepVision AI: An Intelligent Question Paper Analysis System Using NLP and Semantic Similarity for Exam Preparation" International Journal of Advanced Multidisciplinary Research and Educational Development, V2(2): Page(335-339) Mar-Apr 2026. ISSN: 3107-6513. www.ijamred.com. Published by Scientific and Academic Research Publishing.