Classification of Bengali Questions Towards a Factoid Question Answering System

This paper demonstrates question classification for Bengali language questions which is the first step towards developing a closed domain based Bengali factoid question answering system. Question classification is an influential step towards a question answering system on which the performance of the whole system depends. We have used machine learning algorithms Stochastic Gradient Descent(SGD), Decision Tree(DT), Support Vector Machine(SVM) and Naive Bayes(NB) for the question classification process where Support Vector Machine(SVM) with linear kernel performs the best providing the accuracy 90.6% over five coarse-grained categories.