TY - GEN
T1 - Automatic question generation system for english reading comprehension
AU - Fung, Yin Chun
AU - Kwok, Jason Chun Wai
AU - Lee, Lap Kei
AU - Chui, Kwok Tai
AU - U, Leong Hou
N1 - Publisher Copyright:
© 2020, Springer Nature Singapore Pte Ltd.
PY - 2020
Y1 - 2020
N2 - This paper presents a web-based automatic question generation (AQG) system to generate reading comprehension questions and multiple-choice (MC) questions on grammar from a given English text. Such system saves teachers’ time on setting questions and facilitates students and their parents to prepare self-learning exercises. Our web-based system can automatically generate Wh-questions (i.e., what, who, when, where, why, and how) and MC grammar questions of selected sentences. Wh-questions can also be generated from user-specified answer phrases. The generation of Wh-questions exploits the pre-trained natural language understanding model, Text-To-Text Transfer Transformer (T5), and an adapted version of the SQuAD 2.0 machine reading comprehension dataset. The generation of MC questions involves identifying regular verbs in a text and using the verb’s lexemes as the answer choices. Our system takes an average time of about 1 s to generate a Wh-question and it generates a MC question almost instantly. User evaluation indicated that our system is easy-to-use and satisfactory in usefulness, usability, and quality, revealing the effectiveness of our system for teachers and parents.
AB - This paper presents a web-based automatic question generation (AQG) system to generate reading comprehension questions and multiple-choice (MC) questions on grammar from a given English text. Such system saves teachers’ time on setting questions and facilitates students and their parents to prepare self-learning exercises. Our web-based system can automatically generate Wh-questions (i.e., what, who, when, where, why, and how) and MC grammar questions of selected sentences. Wh-questions can also be generated from user-specified answer phrases. The generation of Wh-questions exploits the pre-trained natural language understanding model, Text-To-Text Transfer Transformer (T5), and an adapted version of the SQuAD 2.0 machine reading comprehension dataset. The generation of MC questions involves identifying regular verbs in a text and using the verb’s lexemes as the answer choices. Our system takes an average time of about 1 s to generate a Wh-question and it generates a MC question almost instantly. User evaluation indicated that our system is easy-to-use and satisfactory in usefulness, usability, and quality, revealing the effectiveness of our system for teachers and parents.
KW - Automatic question generation
KW - English reading comprehension
KW - Multiple-choice questions
KW - Natural language processing
KW - Wh-questions
UR - http://www.scopus.com/inward/record.url?scp=85098249352&partnerID=8YFLogxK
U2 - 10.1007/978-981-33-4594-2_12
DO - 10.1007/978-981-33-4594-2_12
M3 - Conference contribution
AN - SCOPUS:85098249352
SN - 9789813345935
T3 - Communications in Computer and Information Science
SP - 136
EP - 146
BT - Technology in Education. Innovations for Online Teaching and Learning - 5th International Conference, ICTE 2020, Revised Selected Papers
A2 - Lee, Lap-Kei
A2 - U, Leong Hou
A2 - Wang, Fu Lee
A2 - Cheung, Simon K.
A2 - Au, Oliver
A2 - Li, Kam Cheong
T2 - 5th International Conference on Technology in Education, ICTE 2020
Y2 - 19 August 2020 through 22 August 2020
ER -