Automated Grading System Using NLP For Student Exams

The advancement of Natural Language Processing (NLP) has significantly influenced various educational technologies. This system explores the development of an NLP-based automated grading system that can process handwritten answer sheets uploaded by students. The system utilizes Optical Character Recognition OCR techniques to convert handwritten responses into machine-readable text, followed by NLP algorithms to evaluate the syntactic and semantic accuracy of answers. The system further calculates individual marks, assigns grades based on performance, and ranks students accordingly. This system aims to address the challenges of manual evaluation, such as time consumption, human bias, and scalability, especially in institutions handling large volumes of answer sheets. The automated systems offer efficiency, fairness and scalability. This paper examines existing methodologies including rule-based, machine learning, and deep learning models, with a focus on their application in syntactic and semantic evaluation. Various similarity measures such as cosine similarity, Jaccard index, and semantic embeddings and architectures including BERT, LSTM, and transformer-based models are analysed. This system demonstrates how combining OCR and NLP can revolutionize academic assessment by making it faster, fairer, and more consistent.