Automatic Question Classifier

Abstract

Bloom’s Taxonomy is an effective tool to provide a common language for teachers to discuss and exchange learning and assessment methods. It is a classification system used to define and distinguish different levels of human cognition–i.e., thinking, learning, and understanding. Currently, academicians identify Bloom’s taxonomy cognitive level manually, but that is a tedious and a time-consuming task. Several studies have been carried out to automate this task, but most of them focus on keywords which can often overlap and fail to understand semantic structure of sentences. This paper aims to propose a question classification model which uses an ensemble approach by combining support vector machines, k-Neural Networks and Naive Bayes algorithm. Feature extraction is done using TFPOS-IDF to assign more weights to verbs in questions and convert text into a matrix (or vector) of features. The ensemble then applies a hard voting algorithm on the output of above classifiers. The application then helps academicians generate different question papers in accordance with Bloom’s Taxonomy.