Corpus-based Automatic Generation of Multiple Choice Questions for College English Tests

At present multiple choice questions as a testing item are adopted widely in College English tests, and play a significant role in assessing students' English proficiency. However, in practice the design of multiple choice questions is also difficult in many aspects, especially with regard to the selection of the question that serves as its stem and the choices that need to be both interfering and exclusive. This paper introduces a method that can generate the stem question automatically based on an English-Chinese parallel corpus in which sentences that meet the designated criteria input by test question makers are extracted on the condition of complex considerations, such as sentence length, word choice and syntactic structure. Although the multiple choice questions generated by our system still need human involvement for correction and consideration, our work will surely provide valuable assistance to teachers who are designing multiple choice questions so that their efficiency will be improved significantly.