MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation

Wenge Liu; Jianheng Tang; Yi Cheng; Wenjie Li; Yefeng Zheng; Xiaodan Liang

doi:10.1007/978-3-031-17120-8_35

MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation

Wenge Liu, Jianheng Tang, Yi Cheng, Wenjie Li, Yefeng Zheng, Xiaodan Liang^*

^*Corresponding author for this work

Research output: Chapter in Book/Conference Proceeding/Report › Conference Paper published in a book › peer-review

27 Citations (Scopus)

Abstract

Medical dialogue systems interact with patients to collect symptoms and provide treatment advice. In this task, medical entities (e.g., diseases, symptoms, and medicines) are the most central part of the dialogues. However, existing datasets either do not provide entity annotation or are too small in scale. In this paper, we present MedDG, an entity-centric medical dialogue dataset, where medical entities are annotated with the help of domain experts. It consists of 17,864 Chinese dialogues, 385,951 utterances, and 217,205 entities, at least one magnitude larger than existing entity-annotated datasets. Based on MedDG, we conduct preliminary research on entity-aware medical dialogue generation by implementing several benchmark models. Extensive experiments show that the entity-aware adaptions on the generation models consistently enhance the response quality but there still remains a large space of improvement for future research. The codes and the dataset are released at https://github.com/lwgkzl/MedDG.

Original language	English
Title of host publication	Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings
Editors	Wei Lu, Shujian Huang, Yu Hong, Xiabing Zhou
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	447-459
Number of pages	13
ISBN (Print)	9783031171192
DOIs	https://doi.org/10.1007/978-3-031-17120-8_35
Publication status	Published - 2022
Externally published	Yes
Event	11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022 - Guilin, China Duration: 24 Sept 2022 → 25 Sept 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13551 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022
Country/Territory	China
City	Guilin
Period	24/09/22 → 25/09/22

Bibliographical note

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Access to Document

10.1007/978-3-031-17120-8_35

Cite this

Liu, W., Tang, J., Cheng, Y., Li, W., Zheng, Y., & Liang, X. (2022). MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation. In W. Lu, S. Huang, Y. Hong, & X. Zhou (Eds.), Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings (pp. 447-459). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13551 LNAI). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-17120-8_35

Liu, Wenge ; Tang, Jianheng ; Cheng, Yi et al. / MedDG : An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation. Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. editor / Wei Lu ; Shujian Huang ; Yu Hong ; Xiabing Zhou. Springer Science and Business Media Deutschland GmbH, 2022. pp. 447-459 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{b6c079eb91d641cea55e8292f5af0348,

title = "MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation",

abstract = "Medical dialogue systems interact with patients to collect symptoms and provide treatment advice. In this task, medical entities (e.g., diseases, symptoms, and medicines) are the most central part of the dialogues. However, existing datasets either do not provide entity annotation or are too small in scale. In this paper, we present MedDG, an entity-centric medical dialogue dataset, where medical entities are annotated with the help of domain experts. It consists of 17,864 Chinese dialogues, 385,951 utterances, and 217,205 entities, at least one magnitude larger than existing entity-annotated datasets. Based on MedDG, we conduct preliminary research on entity-aware medical dialogue generation by implementing several benchmark models. Extensive experiments show that the entity-aware adaptions on the generation models consistently enhance the response quality but there still remains a large space of improvement for future research. The codes and the dataset are released at https://github.com/lwgkzl/MedDG.",

author = "Wenge Liu and Jianheng Tang and Yi Cheng and Wenjie Li and Yefeng Zheng and Xiaodan Liang",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022 ; Conference date: 24-09-2022 Through 25-09-2022",

year = "2022",

doi = "10.1007/978-3-031-17120-8\_35",

language = "English",

isbn = "9783031171192",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "447--459",

editor = "Wei Lu and Shujian Huang and Yu Hong and Xiabing Zhou",

booktitle = "Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings",

}

Liu, W, Tang, J, Cheng, Y, Li, W, Zheng, Y & Liang, X 2022, MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation. in W Lu, S Huang, Y Hong & X Zhou (eds), Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13551 LNAI, Springer Science and Business Media Deutschland GmbH, pp. 447-459, 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022, Guilin, China, 24/09/22. https://doi.org/10.1007/978-3-031-17120-8_35

MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation. / Liu, Wenge; Tang, Jianheng; Cheng, Yi et al.
Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. ed. / Wei Lu; Shujian Huang; Yu Hong; Xiabing Zhou. Springer Science and Business Media Deutschland GmbH, 2022. p. 447-459 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13551 LNAI).

Research output: Chapter in Book/Conference Proceeding/Report › Conference Paper published in a book › peer-review

TY - GEN

T1 - MedDG

T2 - 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022

AU - Liu, Wenge

AU - Tang, Jianheng

AU - Cheng, Yi

AU - Li, Wenjie

AU - Zheng, Yefeng

AU - Liang, Xiaodan

PY - 2022

Y1 - 2022

N2 - Medical dialogue systems interact with patients to collect symptoms and provide treatment advice. In this task, medical entities (e.g., diseases, symptoms, and medicines) are the most central part of the dialogues. However, existing datasets either do not provide entity annotation or are too small in scale. In this paper, we present MedDG, an entity-centric medical dialogue dataset, where medical entities are annotated with the help of domain experts. It consists of 17,864 Chinese dialogues, 385,951 utterances, and 217,205 entities, at least one magnitude larger than existing entity-annotated datasets. Based on MedDG, we conduct preliminary research on entity-aware medical dialogue generation by implementing several benchmark models. Extensive experiments show that the entity-aware adaptions on the generation models consistently enhance the response quality but there still remains a large space of improvement for future research. The codes and the dataset are released at https://github.com/lwgkzl/MedDG.

AB - Medical dialogue systems interact with patients to collect symptoms and provide treatment advice. In this task, medical entities (e.g., diseases, symptoms, and medicines) are the most central part of the dialogues. However, existing datasets either do not provide entity annotation or are too small in scale. In this paper, we present MedDG, an entity-centric medical dialogue dataset, where medical entities are annotated with the help of domain experts. It consists of 17,864 Chinese dialogues, 385,951 utterances, and 217,205 entities, at least one magnitude larger than existing entity-annotated datasets. Based on MedDG, we conduct preliminary research on entity-aware medical dialogue generation by implementing several benchmark models. Extensive experiments show that the entity-aware adaptions on the generation models consistently enhance the response quality but there still remains a large space of improvement for future research. The codes and the dataset are released at https://github.com/lwgkzl/MedDG.

UR - https://www.scopus.com/pages/publications/85140483225

U2 - 10.1007/978-3-031-17120-8_35

DO - 10.1007/978-3-031-17120-8_35

M3 - Conference Paper published in a book

AN - SCOPUS:85140483225

SN - 9783031171192

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 447

EP - 459

BT - Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings

A2 - Lu, Wei

A2 - Huang, Shujian

A2 - Hong, Yu

A2 - Zhou, Xiabing

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 24 September 2022 through 25 September 2022

ER -

Liu W, Tang J, Cheng Y, Li W, Zheng Y, Liang X. MedDG: An Entity-Centric Medical Consultation Dataset for Entity-Aware Medical Dialogue Generation. In Lu W, Huang S, Hong Y, Zhou X, editors, Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 447-459. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-17120-8_35