Question Generation using POS Tagging

In education, human-generated questions are commonly utilized to evaluate a student’s comprehension of a text. However, creating a computer program to do this is challenging due to the complexities of human language. Many existing methods rely on expensive tools for accuracy. Considering our research objectives, we suggest a different way of doing things that uses machine learning and pattern matching based on parts of speech to generate knowledge-testing questions from English text without the need for large semantic tools. This paper demonstrates the utilization of the Stanford Question Answering Dataset to improve question-answering performance on textual data extracted from PDF and TXT files. The research findings demonstrate the effective use of question-answering models, trained on SQuAD, to process and extract valuable information from PDF and TXT documents.