Automatic Multiple-Choice Question and Answer (MCQA) Generation Using Deep Learning Model

Abstract

The ability to extract useful information from very large data sources is known as data mining. Nowadays, semantic web is increasing its user attention in several research inventions especially in the field of artificial intelligence and neural networks. As the world is using AI and ML for all types of applications, there is a huge demand for these domains. In this current work, we are taking MCQA as the main problem statement and trying to propose a novel model which can automatically generate MCQA with more accuracy and less time delay. All the primitive models are either manual approach or ML-based generation models which are not completely successful in designing MCQS without any errors. As the data size increases, the primitive methods try to get a problem like overfitting, and all the data is not going to generate MCQA properly. Hence in this current work, we try to design a deep learning model for generating multiple-choice questions and answers automatically for a given phrase or paragraph with more accuracy and efficient way. By conducting various experiments on our proposed model by taking sample Stanford question answering dataset, which consists of more than one lakh question, some answered and some are not answered. The proposed model is trained on this dataset and finally able to generate MCQA for any type of context which is given by the end-user.