Toward automatic generation of multiple-choice questions in an educational context

automatic question generation

Kang, Zheng ; Zhai ; ChengXiang

关键词: automatic question generation;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/88089/KANG-THESIS-2015.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Traditionally, textbooks often include thought-provoking questions at the beginning or at the end of a piece of text. These questions help readers to self-evaluate their comprehension of the text, and to reinforce their knowledge.The Internet has given people access to much more information than ever available before. With it, we also witness the blossom of educational websites and articles online. While the emergence of these websites and content themselves are great for child and adult learners alike, most of them are not accompanied by such reviewing questions traditionally available in textbooks. Without these questions, readers cannot easily assess their understanding, and forget what they learn in a shorter time span. This makes for a less satisfactory learning experience.It is impractical to generate such questions manually due to the sheer amount of articles available online. Therefore, we propose a method of generating these questions automatically without human intervention. We first select the most important sentences in an article. For each sentence, we automatically choose the most important phrase in it, and blank out that phrase. Finally, we attempt to create distracting answers for the phrase blanked out, making a multiple-choice question out of the sentence.We believe this is a step forward to creating a more enjoyable learning environment for life-long learners and all Internet users.