TY - GEN
T1 - Knowledge Graph Construction for Healthcare Services in Traditional Chinese Medicine
AU - Yi, Zhiwei
AU - Zhang, Bolin
AU - Deng, Xingpeng
AU - Wang, Jiahao
AU - Tu, Zhiying
AU - Chu, Dianhui
AU - Hu, Xin
AU - Ding, Deqiong
AU - Guan, Yong
AU - Sun, Zhao
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
PY - 2023
Y1 - 2023
N2 - Traditional Chinese medicine (TCM) is a bright pearl in the treasure house of healthcare applications that has attracted increasing attention due to its huge applying potential, especially in the prevention and intervention of COVID-19 Pandemic. Such applications for healthcare decision-making are powerful tools to help provide actionable and explainable medical services to patients, but they are a knowledge-driven system and rely on knowledge graphs. However, most of TCM-related materials and guidebooks are preserved in the form of documents, lacking structured information and conceptual knowledge. To facility the study of domain-specific knowledge graphs in TCM, we define the ontology of knowledge graph in TCM with 29 types of entities and 32 types of relations, and then annotate a high-quality dataset (TCM-ERE) for Entity and Relation Extraction (E &RE) aligning with the concepts of the TCM-ontology. More than 40% of relations can only be inferred from multiple sentences in TCM-ERE, thus it can also be used for Chinese document-level E &RE research. The baseline models trained on the TCM-ERE are used to extract fact triples from TCM medical records for the enriching scale of the TCM-related knowledge graph (TCM-KG). TCM-ERE, TCM-KG and the baseline models are publicly available at https://gitee.com/yi_zhi_wei/acup1.git.
AB - Traditional Chinese medicine (TCM) is a bright pearl in the treasure house of healthcare applications that has attracted increasing attention due to its huge applying potential, especially in the prevention and intervention of COVID-19 Pandemic. Such applications for healthcare decision-making are powerful tools to help provide actionable and explainable medical services to patients, but they are a knowledge-driven system and rely on knowledge graphs. However, most of TCM-related materials and guidebooks are preserved in the form of documents, lacking structured information and conceptual knowledge. To facility the study of domain-specific knowledge graphs in TCM, we define the ontology of knowledge graph in TCM with 29 types of entities and 32 types of relations, and then annotate a high-quality dataset (TCM-ERE) for Entity and Relation Extraction (E &RE) aligning with the concepts of the TCM-ontology. More than 40% of relations can only be inferred from multiple sentences in TCM-ERE, thus it can also be used for Chinese document-level E &RE research. The baseline models trained on the TCM-ERE are used to extract fact triples from TCM medical records for the enriching scale of the TCM-related knowledge graph (TCM-KG). TCM-ERE, TCM-KG and the baseline models are publicly available at https://gitee.com/yi_zhi_wei/acup1.git.
KW - Healthcare services
KW - Knowledge extraction
KW - Knowledge graph
KW - Traditional Chinese medicine
UR - https://www.scopus.com/pages/publications/85172656687
U2 - 10.1007/978-981-99-4402-6_23
DO - 10.1007/978-981-99-4402-6_23
M3 - 会议稿件
AN - SCOPUS:85172656687
SN - 9789819944019
T3 - Communications in Computer and Information Science
SP - 321
EP - 335
BT - Service Science - CCF 16th International Conference, ICSS 2023, Revised Selected Papers
A2 - Wang, Zhongjie
A2 - Xu, Hanchuan
A2 - Wang, Shangguang
PB - Springer Science and Business Media Deutschland GmbH
T2 - 16th International Conference on Service Science, ICSS 2023
Y2 - 13 May 2023 through 14 May 2023
ER -