International Journal of Computer Theory and Engineering

Abstract

Consider generating quizzes for a knowledge of a category in order to make use of learning. Particularly, if we realized generating quizzes automatically by using Wikipedia, which is the vast treasure trove of knowledge, it might greatly contribute to learners. In this study, we propose a system that generates fill-in-the-blanks-type quizzes about a category consisting of the elements of the fixed number. To achieve this, the system calculates the tf-idf values for each word in retrieved documents from Wikipedia, and the degree of importance of sentences, then, outputs sentences in order of the degree of importance. In order to seek the most appropriate function for the degree of importance of a sentence, for some sample documents, we compare the correlation coefficient between the ranking obtained by human cooperators and the ranking generated by each of fifteen candidate functions.