Key points are not available for this paper at this time.
Abstract The world is moving towards computerization. Manually checking the answer sheets takes a lot of time and effort from the schoolteachers and college professors. To address this challenge, our project aims to streamline the evaluation process by converting handwritten student responses into digital text and comparing them with predetermined model answers provided by educators. This is made possible through the integration of cutting-edge technologies such as optical character recognition (OCR), natural language processing (NLP), and machine learning algorithms. By the utilization of advanced BERT (Bidirectional Encoder Representations from Transformers) models and cosine similarity algorithms, our system ensures accuracy and efficiency when evaluating student answers. Rather than focusing on answer length, the project's main goal is to optimize mark distribution based on key terms. This will save educators time and effort while advancing a fair and uniform evaluation process. Additionally, this approach helps students understand concepts more clearly and motivates them to give exact and accurate answers, which helps produce results that are fair and equal.
Kulkarni et al. (Fri,) studied this question.