Automated Question Generation Methods for Intelligent English Learning Systems and its Evaluation

This paper presents methods of automated generation of questions about the contents of an English story. The methods are used for realizing adaptive Question and Answer functions which give suitable questions to learners of different understanding states. To realize adaptive Question and Answer, our methods generate five kinds of questions; to ask about the content of one sentence, to use antonyms and synonyms, to use modifiers appeared in plural sentences, to ask the contents represented by plural sentences by using a relative pronoun, and to ask time and space relationships. Our methods cover 80% of questions in collection books of problems for novices. The result of an experiment shows that the question generation module generates 1977 question sentences from 301 sentences in textbooks for Japanese junior high school students, and 1848 of the generated question sentences are semantically competent.