Machine Learning Techniques for Descriptive Answer Evaluation: A Comprehensive Survey

This research study intends to assist the examination board in evaluating the responses based on the level of question difficulty determined by the board. As an initial step, this study reviews numerous approaches and techniques for evaluating descriptive answers using machine learning. In order to evaluate descriptive responses automatically, this research study proposes a novel approach that makes use of various machine learning, natural language processing, and tools like WordNet, Word Mover's Distance (WMD), cosine similarity, Latent Semantic Analysis (LSA), Word2vec, LSTM Bi-directional Encoder Representation from Transformer. This analysis has examined 20 articles that were published in the recent years, from 2012 to 2022. Based on this research and investigation, this study concludes LSTM as the optimum method for descriptive answer evaluation using machine learning techniques. With the aim to increase the precision and the efficiency of descriptive answer evaluation, research works have been actively conducted to develop these methods.