A Method for Automatic Generation of Multiple-Choice Questions Adapted to Students’ Understanding

Abstract

The purpose of this research is to propose a method for generating multiple-choice questions with adapted level of difficulty automatically. Multiple-choice question is suitable for confirming knowledge like the relation of the name of a word and its property, though it is hardly used in order to confirm in-depth knowledge with an analogy. Moreover, it is often used for the large-scale examination because of its high objectivity and its easiness of mechanical grading. We aim to materialize a question generator for students who will take qualifying / certificate examinations to experience many questions adapted to students’ understanding. As multiple-choice questions can be based on the fixed forms by question sentences, it is easy to generate questions mechanically using a database for term about “computer” domain. In addition, it is possible to control the difficulty of multiple-choice questions by selecting alternatives. In this paper, a method to generate multiple-choice questions, a system to generate questions automatically, and an evaluation of this system are discussed.