Construction of Terminology Semantic Knowledge Base Based on Conceptual Structure and Distributed Representation

doi:10.12339/j.issn.1673-8578.2023.04.001

Abstract

Abstract:

The construction of lexical semantic knowledge base is a basic task in natural language processing and plays an important role in various subtasks of natural language processing. This paper proposes an automatic generation method of term concept KDML representation for compound terms. The method is based on the hierarchical structure of concepts, uses distributed representation method to represent concepts and term definition text, and performs concept disambiguation according to the semantic distance between concepts and terms definition text. The method can learn the semantic roles between concepts through K-nearest neighbor algorithm, and generates term concept representation according to KDML syntax rules. Our experimental results showed that the accuracy rate of the first sememe was 96%, and the F1 values of the total sememe, semantic role and total triple were 91.92%, 78.9% and 73.41% respectively.

Key words: distributed representation, semantic disambiguation, HowNet, construction of terminology semantic knowledge base

WANG Peiyan, LI Linna, SHEN Sijia. Construction of Terminology Semantic Knowledge Base Based on Conceptual Structure and Distributed Representation[J]. China Terminology, 2023, 25(4): 3-11.

Figures/Tables 12

References 21

[1]	董振东, 董强. KDML:知网知识系统描述语言[Z/OL]. [2023-03-07]. http://www.keenage.com/zhiwang/c_zhiwang.html.
[2]	董振东, 董强. 面向信息处理的词汇语义研究中的若干问题[J]. 语言文字应用, 2001(3): 27-32.
[3]	LIU Q, LI S J. Word similarity computing based on how-net[J]. Comput. Linguistics Chin. Lang. Process, 2002, 7(2):59-76.
[4]	FU X H, GUO L, GAO Y Y, et al. Multi-aspect sentiment analysis for Chinese online social reviews based on topic modeling and hownet lexicon[J]. Knowledge-Based Systems, 2013, 37(2):186-195. doi: 10.1016/j.knosys.2012.08.003 URL
[5]	NIU Y L, XIE R B, YUAN X C, et al. Improved word representation learning with sememes[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, 2017: 2049-2058.
[6]	XIE R B, YUAN X C, LIU Z Y, et al. Lexical sememe prediction via word embeddings and matrix factorization[C]// IJCAI, 2017: 4200-4206.
[7]	GU Y H, YAN J, ZHU H, et al. Language modeling with sparse product of sememe experts[J/OL]. arXiv preprint(2018-10-29)[2023-03-07]. https://arxiv.org/abs/1810.12387v1.
[8]	QI F C, YANG C H, LIU Z Y, et al. Openhownet: An open sememe-based lexical knowledge base[J/OL]. arXiv preprint(2019-01-28)[2023-03-07]. https://doi.org/10.48550/arXiv.1901.09957.
[9]	DONG Z D, DONG Q. Hownet and the computation of meaning[M]. Singapore: World Scientific Publishing Company, 2006.
[10]	KDML: 知网知识系统描述语言[Z/OL]. [2023-03-07]. http://www.keenage.com/TheoryandpracticeofHowNet/07.pdf.
[11]	王裴岩, 张桂平, 蔡东风. 航空术语语义知识库ATHowNet的构建[J]. 中国科技术语, 2021, 23(3): 26-32. doi: 10.12339/j.issn.1673-8578.2021.03.004
[12]	LIU L Q, WANG Q L, LI Y. Improved Chinese sentence semantic similarity calculation method based on multi-feature fusion[J]. J. Adv. Comput. Intell. Intell. Informatics, 2021, 25(4): 442-449.
[13]	BORDES A, WESTON J, COLLOBERT R, et al. Lear-ning structured embeddings of knowledge bases[C]// Proceedings of the AAAI conference on artificial intelligence, 2011, 25(1): 301-306.
[14]	WU P C, HOI S C-H, ZHAO P L, et al. Mining social images with distance metric learning for automated image tagging[C]// Proceedings of the fourth ACM international conference on Web search and data mining, 2011: 197-206.
[15]	MIKOLOV T, CHEN K, CORRADO G, et al. Efficient estimation of word representations in vector space[J/OL]. arXiv preprint(2013-09-07) [2023-03-07]. https://doi.org/10.48550/arXiv.1301.3781.
[16]	余军成. 概念及其内涵和外延述评[J]. 安徽文学(下半月), 2007 (1): 121.
[17]	孙寰. 汉源术语的本质及其特征[J]. 中国科技术语, 2023, 25 (1): 10-19. doi: 10.12339/j.issn.1673-8578.2023.01.002
[18]	李江华, 时鹏, 胡长军. 一种适用于复合术语的本体概念学习方法[J]. 计算机科学, 2013, 40 (5): 168-172.
[19]	张绍友. 也论概念内涵与外延的关系:兼与樊明亚先生商榷[C]// 逻辑研究文集: 中国逻辑学会第六次代表大会暨学术讨论会论文集, 2000: 213-221.
[20]	戚晓杰. 关于种属概念词语法结构类型问题的思考[J]. 汉语学习, 2022 (1): 37-46.
[21]	赵超丽, 王裴岩, 蔡东风. 面向航空领域 HowNet 构建的术语内部动态角色标注[J]. 计算机应用与软件, 2021, 38(3): 163-168,209.

术语	概念表示
橡胶减震垫	DEF={part\|部件:{weaken\|削弱:patient={shiver\|颤动},material={material\|材料}},whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
义原
part\|部件	shiver\|颤动
implement\|器具	material\|材料
inlay\|嵌入	tool\|用具
weaken\|削弱	shiver\|驱动
语义角色
whole(整体是...)	instrument(用具是...)
patient(受事是...)	material(材料是...)

术语	概念表示
橡胶减震垫	DEF={part\|部件:{weaken\|削弱:patient={shiver\|颤动},material={material\|材料}},whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
义原
part\|部件	shiver\|颤动
implement\|器具	material\|材料
inlay\|嵌入	tool\|用具
weaken\|削弱	shiver\|驱动
语义角色
whole(整体是...)	instrument(用具是...)
patient(受事是...)	material(材料是...)

术语	概念表示
橡胶减震垫	DEF={part\|部件:{weaken\|削弱:patient={shiver\|颤动},material={material\|材料}},whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
子词(概念)	概念表示
垫	{part\|部件:whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
减震	{weaken\|削弱:patient={shiver\|颤动}
橡胶	{material\|材料}
概念关系(语义角色)
垫	修饰(省略)	减震
垫	material(材料是...)	橡胶

术语	概念表示
橡胶减震垫	DEF={part\|部件:{weaken\|削弱:patient={shiver\|颤动},material={material\|材料}},whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
子词(概念)	概念表示
垫	{part\|部件:whole={implement\|器具},{inlay\|嵌入:instrument={~}}}
减震	{weaken\|削弱:patient={shiver\|颤动}
橡胶	{material\|材料}
概念关系(语义角色)
垫	修饰(省略)	减震
垫	material(材料是...)	橡胶

子词		概念表示	说明
垫	1	{part\|部件:whole={implement\|器具},{inlay\|嵌入:instrument={~}}}	名词,作为部件不能独立使用的垫,如“衬垫”
	2	{tool\|用具:{inlay\|嵌入:instrument={~}}}	名词,能作为用具单独使用的垫,如“气垫”
	3	{pay\|付}	动词,支付
	4	{inlay\|嵌入}	动词,垫入
减震	1	{weaken\|削弱:patient={shiver\|颤动}}	动词
橡胶	1	{material\|材料}	名词