A Lecture Centric Automated Distractor Generation for Post-Graduate Software Engineering Courses

Abstract

Critical circumstances, natural disasters or pandemics like COVID 19 gave rise to the wide applicability of E-learning into education system. Efficient and fair online assessment is very important to utilize the inevitable benefits of E-learning.. In order to make it efficient, the trend of assessment has shifted from the subjective type to the objective type assessments which is mainly based on Multiple Choice Questions (MCQs), generation of MCQs is a tedious, tiresome and time consuming task. To cater this dire need, this study proposes an automated Multiple Choice Question (MCQ) generation by utilizing state of the art transformer based model T5 for the task of question generation and a lexicon based approach Sense2vec for the task of distractor generation. It also presented a domain specific lecture text based test data for performing evaluation on the task of domain specific lecture text based MCQ generation.