Machine Learning Based Automatic Answer Script Evaluation Using Artificial Neural Network

In the current era of Automation, there is a growing need for automated systems to evaluate descriptive answer sheets more effectively. Manual grading has become labor intensive, time consuming and prone to make errors. While automated systems exist for objective and brief answers, but often lack in descriptive evaluation. The pre-processed system processes the scanned handwritten answer sheet, then using Optical Character Recognition (OCR) it converts the handwritten answer script images to Text. Finally using Natural language processing (NLP) techniques, including tokenization and word embeddings are used to analyze text based on keywords and length criteria provided by the moderator. A Feedforward artificial neural network (ANN) with back-propagation algorithm is used for classifying and recognition of the content. This System mainly focuses on accurate, consistent evaluation and feedback, reducing traditional manual efforts and to improve grading efficiency. The proposed model is providing higher accuracy level compare to existing method.