Few-shot Question Generation for Reading Comprehension

Yin Poon; John S. Y. Lee; Yu Yan Lam; Wing Lam Suen; Elsie Li Chen Ong; Samuel Kai Wah Chu

doi:10.18653/v1/2024.sighan-1.3

Few-shot Question Generation for Reading Comprehension

Yin Poon, John S. Y. Lee, Yu Yan Lam, Wing Lam Suen, Elsie Li Chen Ong, Samuel Kai Wah Chu

Department of Linguistics and Translation

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

1 Citation (Scopus)

33 Downloads (CityUHK Scholars)

Abstract

According to the internationally recognized PIRLS (Progress in International Reading Literacy Study) assessment standards, reading comprehension questions should require not only information retrieval, but also higher-order processes such as inferencing, interpreting and evaluation. However, these kinds of questions are often not available in large quantities for training question generation models. This paper investigates whether pre-trained Large Language Models (LLMs) can produce higher-order questions. Human assessment on a Chinese dataset shows that few-shot LLM prompting generates more usable and higher-order questions than two competitive neural baselines. © 2024 Association for Computational Linguistics.

Original language	English
Title of host publication	Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10)
Publisher	Association for Computational Linguistics
Pages	21-27
ISBN (Print)	9798891761551
DOIs	https://doi.org/10.18653/v1/2024.sighan-1.3
Publication status	Published - Aug 2024
Event	10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10) - Hybrid, Bangkok, Thailand Duration: 16 Aug 2024 → … https://aclanthology.org/2024.sighan-1

Conference

Conference	10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10)
Country/Territory	Thailand
City	Bangkok
Period	16/08/24 → …
Internet address	https://aclanthology.org/2024.sighan-1

Funding

We gratefully acknowledge support from the eLearning Ancillary Facilities Programme of the Quality Education Fund (Project “Knowledge Overlord - A Self-sustaining AI Game-based Online Platform to Enhance Student’s Literacy Ability and 21st Century Skills”); and from a Teaching Development Grant at City University of Hong Kong (project 6000834).

Publisher's Copyright Statement

This full text is made available under CC-BY 4.0. https://creativecommons.org/licenses/by/4.0/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access Output

10.18653/v1/2024.sighan-1.3

237352185.pdfFinal Published version, 342 KBLicence: CC BY

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inproceedings{a707c017e123408cb1d2375eeeb89675,

title = "Few-shot Question Generation for Reading Comprehension",

abstract = "According to the internationally recognized PIRLS (Progress in International Reading Literacy Study) assessment standards, reading comprehension questions should require not only information retrieval, but also higher-order processes such as inferencing, interpreting and evaluation. However, these kinds of questions are often not available in large quantities for training question generation models. This paper investigates whether pre-trained Large Language Models (LLMs) can produce higher-order questions. Human assessment on a Chinese dataset shows that few-shot LLM prompting generates more usable and higher-order questions than two competitive neural baselines. {\textcopyright} 2024 Association for Computational Linguistics.",

author = "Yin Poon and Lee, {John S. Y.} and Lam, {Yu Yan} and Suen, {Wing Lam} and Ong, {Elsie Li Chen} and Chu, {Samuel Kai Wah}",

year = "2024",

month = aug,

doi = "10.18653/v1/2024.sighan-1.3",

language = "English",

isbn = "9798891761551",

pages = "21--27",

booktitle = "Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10)",

publisher = "Association for Computational Linguistics",

address = "United States",

note = "10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10) ; Conference date: 16-08-2024",

url = "https://aclanthology.org/2024.sighan-1",

}

Poon, Y, Lee, JSY, Lam, YY, Suen, WL, Ong, ELC & Chu, SKW 2024, Few-shot Question Generation for Reading Comprehension. in Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10). Association for Computational Linguistics, pp. 21-27, 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10), Bangkok, Thailand, 16/08/24. https://doi.org/10.18653/v1/2024.sighan-1.3

Few-shot Question Generation for Reading Comprehension. / Poon, Yin; Lee, John S. Y.; Lam, Yu Yan et al.
Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10). Association for Computational Linguistics, 2024. p. 21-27.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Few-shot Question Generation for Reading Comprehension

AU - Poon, Yin

AU - Lee, John S. Y.

AU - Lam, Yu Yan

AU - Suen, Wing Lam

AU - Ong, Elsie Li Chen

AU - Chu, Samuel Kai Wah

PY - 2024/8

Y1 - 2024/8

N2 - According to the internationally recognized PIRLS (Progress in International Reading Literacy Study) assessment standards, reading comprehension questions should require not only information retrieval, but also higher-order processes such as inferencing, interpreting and evaluation. However, these kinds of questions are often not available in large quantities for training question generation models. This paper investigates whether pre-trained Large Language Models (LLMs) can produce higher-order questions. Human assessment on a Chinese dataset shows that few-shot LLM prompting generates more usable and higher-order questions than two competitive neural baselines. © 2024 Association for Computational Linguistics.

AB - According to the internationally recognized PIRLS (Progress in International Reading Literacy Study) assessment standards, reading comprehension questions should require not only information retrieval, but also higher-order processes such as inferencing, interpreting and evaluation. However, these kinds of questions are often not available in large quantities for training question generation models. This paper investigates whether pre-trained Large Language Models (LLMs) can produce higher-order questions. Human assessment on a Chinese dataset shows that few-shot LLM prompting generates more usable and higher-order questions than two competitive neural baselines. © 2024 Association for Computational Linguistics.

UR - http://www.scopus.com/inward/record.url?scp=85204880373&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85204880373&origin=recordpage

U2 - 10.18653/v1/2024.sighan-1.3

DO - 10.18653/v1/2024.sighan-1.3

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9798891761551

SP - 21

EP - 27

BT - Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10)

PB - Association for Computational Linguistics

T2 - 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10)

Y2 - 16 August 2024

ER -

Few-shot Question Generation for Reading Comprehension

Abstract

Conference

Funding

Publisher's Copyright Statement

UN SDGs

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

TDG(CityU): Chatbots as Pedagogical Tools: A Feasibility Study

Cite this

Few-shot Question Generation for Reading Comprehension

Abstract

Conference

Funding

Publisher's Copyright Statement

UN SDGs

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Projects

TDG(CityU): Chatbots as Pedagogical Tools: A Feasibility Study

Cite this