TY - GEN
T1 - Frequently Asked Question Pair Generation for Rule and Regulation Document
AU - Ding, Keyang
AU - Cai, Chenran
AU - Huang, Shijue
AU - Wang, Rui
AU - Wang, Qianlong
AU - Li, Jianxin
AU - Shi, Guozhong
AU - Hu, Feiran
AU - Li, Fengxin
AU - Xu, Ruifeng
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - This paper presents a novel task to generate frequently asked question (FAQ) pairs for the rule and regulation documents. It offers an easy way for customers and employers to quickly gain knowledge of them and provides a potential corpus for question-answering robots. While previous work focuses on web texts (e.g., Wiki), we generate FAQ pairs from the formal and verbose rule and regulation documents, which is significant in real scenarios. To tackle this task, firstly, we carefully design a rules-based method to generate FAQ pairs based on structure information. Then we propose a pipeline framework for FAQ pair generation by deep learning. For experiments, we collect and annotate a Chinese FAQ pair generation dataset from documents of China Merchants Securities Co., Ltd. The results show that our method can generate proper FAQ pairs and achieve competitive performance in both automatic and human evaluation.
AB - This paper presents a novel task to generate frequently asked question (FAQ) pairs for the rule and regulation documents. It offers an easy way for customers and employers to quickly gain knowledge of them and provides a potential corpus for question-answering robots. While previous work focuses on web texts (e.g., Wiki), we generate FAQ pairs from the formal and verbose rule and regulation documents, which is significant in real scenarios. To tackle this task, firstly, we carefully design a rules-based method to generate FAQ pairs based on structure information. Then we propose a pipeline framework for FAQ pair generation by deep learning. For experiments, we collect and annotate a Chinese FAQ pair generation dataset from documents of China Merchants Securities Co., Ltd. The results show that our method can generate proper FAQ pairs and achieve competitive performance in both automatic and human evaluation.
KW - Deep learning
KW - Frequently asked question pair generation
KW - Natural language processing
UR - https://www.scopus.com/pages/publications/85144821988
U2 - 10.1007/978-3-031-23504-7_4
DO - 10.1007/978-3-031-23504-7_4
M3 - 会议稿件
AN - SCOPUS:85144821988
SN - 9783031235030
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 49
EP - 63
BT - Artificial Intelligence and Mobile Services – AIMS 2022 - 11th International Conference, Held as Part of the Services Conference Federation, SCF 2022, Proceedings
A2 - Pan, Xiuqin
A2 - Jin, Ting
A2 - Zhang, Liang-Jie
PB - Springer Science and Business Media Deutschland GmbH
T2 - 11th International Conference on Artificial Intelligence and Mobile Services, AIMS 2022 held as Part of the Services Conference Federation, SCF 2022
Y2 - 10 December 2022 through 14 December 2022
ER -