A Survey: ML-Based Automated Handwriting Analysis and Answer Evaluation

Abstract: In a technologically advancing world, the evaluation of answers should happen rapidly and with greater accuracy. However, unlike objective answers, subjective answers make it difficult for an automated system to evaluate them accurately. This is because subjective answers are hard to evaluate using static content and finding a dynamic capability that caters to content, meaning, order and structure for subjective type answer evaluation is not so easy. This study represents an automated evaluation system for handwritten as well as textual answer sheets making use of ML and NLP for the evaluation. This survey is all about a system that converts the answers written on the answer sheets into their digital text data, then check whether answer of each question is correct or not. This study comprises of various “Machine Learning” algorithm to recognize and digitize text from handwritten forms. It also analyzes the answer of a student based on keyword matching, semantic similarity and correct grammar and according to that it assigns marks for their given answer using various “Machine Learning” techniques and algorithms. These systems help to minimize biased marking scheme and promotes fair grading. Also, ensuring consistent evaluation and less human work. An overview has been provided, which includes its evolution and effectiveness of various Machine Learning (ML) techniques to improve “Subjective answer evaluation systems”.

Keywords: Optical Character Recognition (OCR), Convolutional Neural Networks (CNN), Machine learning (ML), Natural Language Processing (NLP), Large Language Models (LLM), Subjective Answer Assessment.