液氧液氢发动机领域知识图谱构建与应用

陈潇萍; 王剑锋; 张虹; 胡庆杰

doi:10.13700/j.bh.1001-5965.2022.0333

液氧液氢发动机领域知识图谱构建与应用

doi: 10.13700/j.bh.1001-5965.2022.0333

1.
北京航天动力研究所，北京 100076
2.
中国船舶科学研究中心，无锡 214082

详细信息

通讯作者:
E-mail：chenxp@calt11.cn

中图分类号: TP391.1；TP18；V434
计量
- 文章访问数: 609
- HTML全文浏览量: 93
- PDF下载量: 30
- 被引次数: 0
出版历程
- 收稿日期: 2022-05-07
- 录用日期: 2022-08-07
- 网络出版日期: 2022-10-31
- 整期出版日期: 2024-03-27

Construction and application of knowledge graph in LOX/LH2 engine domain

1.
Beijing Aerospace Propulsion Institute，Beijing 100076，China
2.
China Ship Scientific Research Center，Wuxi 214082，China

More Information

Corresponding author: E-mail：chenxp@calt11.cn

摘要

摘要:
液氧液氢发动机作为航天领域的关键分系统，为有效开展其研制过程的智能化转型工作，针对液氧液氢发动机领域构建知识图谱，在储备领域知识的同时，提高科研生产人才的培养能力。针对液氧液氢发动机领域特点，对领域语料标注、领域知识识别、实体关系识别3个方面进行了研究，在此基础上进行了领域知识图谱的构建，并从领域知识搜索、知识推荐、探索式分析3个角度梳理了其业务应用模式。形成了液氧液氢发动机领域知识体系，并提出构建方法和应用模式。研究结果为航天领域智能化转型提供了参考。
- 液氧液氢 /
- 发动机 /
- 领域知识图谱 /
- 知识实体 /
- 关系抽取
Abstract:
As a crucial component of the aerospace industry, we construct the domain knowledge graph in the domain of liquid oxygen and liquid hydrogen engines in order to maintain domain knowledge and efficiently enhance the training capacity of scientific research and production talents. According to the characteristics of this domain, three aspects of domain corpus labeling, domain entity recognition, and entity relationship recognition are studied. Based on the research results, the construction of a domain knowledge graph is carried out, and the application mode from three perspectives is sorted out: domain knowledge search, knowledge recommendation, and exploratory analysis. Ultimately, the building methods and application modes are proposed, and the knowledge system in the field of liquid oxygen and liquid hydrogen engines is developed. These materials serve as a guide for the intelligent transformation of the aerospace sector.
- liquid oxygen and liquid hydrogen /
- engine /
- domain knowledge graph /
- knowledge entity /
- relation extraction

HTML全文

图 1 知识图谱应用

Figure 1. Knowledge graph applications

下载: 全尺寸图片幻灯片

图 2 液氧液氢发动机领域知识图谱技术架构

Figure 2. Liquid oxygen and liquid hydrogen engine domain knowledge graph technical architecture

下载: 全尺寸图片幻灯片

图 3 液氧液氢发动机领域知识图谱逻辑架构

Figure 3. Logical architecture of knowledge graph for liquid oxygen and liquid hydrogen engine domain

下载: 全尺寸图片幻灯片

图 4 领域语料标注结果

Figure 4. Domain corpus labelling results

下载: 全尺寸图片幻灯片

图 5 关系语料标注结果

Figure 5. Relational corpus labelling results

下载: 全尺寸图片幻灯片

图 6 3层模型工作流程

Figure 6. Three-layer model workflow

下载: 全尺寸图片幻灯片

图 7 广义后缀树算法

Figure 7. Generalised suffix tree algorithm

下载: 全尺寸图片幻灯片

图 8 t时刻下的BiLSTM示例

Figure 8. Example of BiLSTM at moment t

下载: 全尺寸图片幻灯片

图 9 实体融合过程的决策策略

Figure 9. Entity integration strategy

下载: 全尺寸图片幻灯片

图 10 知识应用系统的主要功能界面

Figure 10. Main functional interface of the knowledge application system

下载: 全尺寸图片幻灯片

图 11 知识图谱检索功能界面

Figure 11. Knowledge graph retrieval function interface

下载: 全尺寸图片幻灯片

图 12 知识实体识别实验结果对比图

Figure 12. Comparison of knowledge entity recognition experimental results

下载: 全尺寸图片幻灯片

图 13 实体关系识别实验结果

Figure 13. Results of the entity relationship recognition experiment

下载: 全尺寸图片幻灯片

图 14 A_N效果对比

Figure 14. A_N effect comparison

下载: 全尺寸图片幻灯片

表 1 主要扩展属性

Table 1. Main extended attributes

概念类型	属性用途	基本构成
文档知识	描述获取该文档的方式，及其内容信息	文档标题、来源、等级、业务方向、存储路径、文档内容
推进剂	描述推进剂的主要成分	成分、配比、储存条件、储存时长、燃烧效率、能量效率
结构	描述发动机的主要结构技术	点火方式、燃烧室、喷嘴、冷却系统、声学系统、排气系统
APP	描述APP的基本信息和使用信息	应用名、软件平台、供应商、运营商、简介、内容语言

下载: 导出CSV

表 2 实验数据情况

Table 2. Experimental data status

数据集	实体数量	实体对	实体关系
训练语料	32 384	834 629	60 947
开发语料	9 102	90 640	5 275
测试语料	11 286	103 809	7 206

下载: 导出CSV

表 3 知识实体识别模型参数设置

Table 3. Knowledge entity recognition model parameter settings

Transformer 层数	向量维度	隐藏层数	Dropout	学习率	批大小	迭代周期
12	256	1024	0.5	3×10⁻⁵	32	100

下载: 导出CSV

表 4 实体关系识别模型参数设置

Table 4. Parameterisation of the entity relationship recognition model

$\theta ({d})$	$\theta ({w})$
5	{3,4}

下载: 导出CSV

表 5 对照参数组

Table 5. Control parameter group

组别	向量维度	隐藏层数	批大小	$ \theta ({\mathrm{dist}})$	$\theta ({\mathrm{wind}})$
A	128	512	16	4	{3,3}
B	128	768	16	4	{3,4}
C	128	1024	16	5	{3,5}
D	256	512	32	6	{4,5}
E	256	768	32	6	{4,6}

下载: 导出CSV

表 6 知识实体识别实验结果

Table 6. Results of knowledge entity recognition experiments

组别	P	R	F₁
A	0.852	0.805	0.828
B	0.894	0.846	0.869
C	0.902	0.869	0.885
D	0.879	0.852	0.865
E	0.910	0.869	0.889
实验组	0.908	0.882	0.895

下载: 导出CSV

表 7 实体关系识别实验结果

Table 7. Results of the entity relationship recognition experiment

组别	P	R	F₁	A₅	A₁₀	A₂₀
A	0.802	0.582	0.675	0.822	0.828	0.830
B	0.789	0.637	0.705	0.797	0.805	0.813
C	0.754	0.783	0.768	0.762	0.785	0.800
D	0.603	0.795	0.686	0.609	0.609	0.615
E	0.585	0.804	0.677	0.627	0.665	0.709
实验组	0.772	0.768	0.770	0.798	0.822	0.835

下载: 导出CSV

参考文献(25)

[1]	PUJARA J, MIAO H, GETOOR L, et al. Ontology-aware partitioning for knowledge graph identification[C]//Proceedings of the Workshop on Automated Knowledge Base Construction. New York: ACM, 2013: 19-24.
[2]	林明. 基于知识图谱的交互关系浏览与分析: 可视化模型与系统实现[D]. 杭州: 浙江大学, 2017. LIN M. Interactive relation exploration and analysis based on knowledge graph: Visualization model and system implementation[D]. Hangzhou: Zhejiang University, 2017(in Chinese).
[3]	王萌, 王昊奋, 李博涵, 等. 新一代知识图谱关键技术综述[J]. 计算机研究与发展, 2022, 59(9): 1947-1965. WANG M, WANG H F, LI B H, et al. Survey of key technologies of new generation knowledge graph[J]. Journal of Computer Research and Development, 2022, 59(9): 1947-1965(in Chinese).
[4]	姚萍, 李坤伟, 张一帆. 知识图谱构建技术综述[J]. 信息系统工程, 2020(5): 121-123. doi: 10.3969/j.issn.1001-2362.2020.05.054 YAO P, LI K W, ZHANG Y F. Summary of knowledge map construction technology[J]. China CIO News, 2020(5): 121-123(in Chinese). doi: 10.3969/j.issn.1001-2362.2020.05.054
[5]	CRANEFIELD S. Networked knowledge representation and exchange using UML and RDF[J/OL]. Journal of Digital Information, [2006-01-24][2022-07-03].
[6]	GUO Y B, PAN Z X, HEFLIN J. LUBM: A benchmark for OWL knowledge base systems[J]. Journal of Web Semantics, 2005, 3(2-3): 158-182. doi: 10.1016/j.websem.2005.06.005
[7]	BOLLACKER K, EVANS C, PARITOSH P, et al. Freebase: A collaboratively created graph database for structuring human knowledge[C]//Proceedings of the ACM SIGMOD International Conference on Management of Data. New York: ACM, 2008: 1247-1250.
[8]	SUCHANEK F M, KASNECI G, WEIKUM G. YAGO: A large ontology from wikipedia and WordNet[J]. Journal of Web Semantics, 2008, 6(3): 203-217. doi: 10.1016/j.websem.2008.06.001
[9]	BIZER C, LEHMANN J, KOBILAROV G, et al. DBpedia—A crystallization point for the web of data[J]. Journal of Web Semantics, 2009, 7(3): 154-165. doi: 10.1016/j.websem.2009.07.002
[10]	MATUSZEK C, WITBROCK M, CABRAL J, et al. An Introduction to the syntax and content of CyC[J/OL]. (2022-07-03)[2006-03].
[11]	NIU X, SUN X R, WANG H F, et al. Zhishi. me—Weaving Chinese linking open data[C]//International Semantic Web Conference. Berlin: Springer, 2011: 205-220.
[12]	XU B, LIANG J Q, XIE C H, et al. CN-DBpedia2: An extraction and verification framework for enriching Chinese encyclopedia knowledge base[J]. Data Intelligence, 2019, 1(3): 271-288. doi: 10.1162/dint_a_00017
[13]	GUPTA S, KENKRE S, TALUKDAR P. CaRe: Open knowledge graph embeddings[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). Stroudsburg: Association for Computational Linguistics, 2019: 378-388.
[14]	中国电子技术标准化研究院. 知识图谱标准化白皮书[R/OL]. (2023-07-03)[2019-09-11]. China Electronics Standardization Institute. Knowledge Graph standardization white paper[R/OL]. (2022-07-03) [2019-09-11].
[15]	刘烨宸, 李华昱. 领域知识图谱研究综述[J]. 计算机系统应用, 2020, 29(6): 1-12. LIU Y C, LI H Y. Survey on domain knowledge graph research[J]. Computer Systems & Applications, 2020, 29(6): 1-12(in Chinese).
[16]	吴建军, 朱晓彬, 程玉强, 等. 液体火箭发动机智能健康监控技术研究进展[J]. 推进技术, 2022, 43(1): 7-19. doi: 10.13675/j.cnki.tjjs.200668 WU J J, ZHU X B, CHENG Y Q, et al. Research progress of intelligent health monitoring technology for liquid-propellant rocket engines[J]. Journal of Propulsion Technology, 2022, 43(1): 7-19(in Chinese). doi: 10.13675/j.cnki.tjjs.200668
[17]	刘丹阳, 方全, 张晓伟, 等. 基于图对比注意力网络的知识图谱补全[J/OL]. 北京航空航天大学学报, 2022, 48(8): 1428-1435. LIU D Y, FANG Q, ZHANG X W, et al. Knowledge graph completion based on graph contrastive attention network[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(8): 1428-1435(in Chinese).
[18]	李涓子, 侯磊. 知识图谱研究综述[J]. 山西大学学报(自然科学版), 2017, 40(3): 454-459. LI J Z, HOU L. Reviews on knowledge graph research[J]. Journal of Shanxi University (Natural Science Edition), 2017, 40(3): 454-459(in Chinese).
[19]	曹高辉, 焦玉英, 成全. 基于凝聚式层次聚类算法的标签聚类研究[J]. 现代图书情报技术, 2008(4): 23-28. CAO G H, JIAO Y Y, CHENG Q. Research on tag cluster based on hierarchical agglomerative clustering algorithm[J]. New Technology of Library and Information Service, 2008(4): 23-28(in Chinese).
[20]	林婧, 何震瀛. 基于广义后缀树结合过滤因子的正则表达式匹配算法[J]. 计算机应用与软件, 2022, 39(1): 266-270. LIN J, HE Z Y. Regular expression matching algorithm based on generalized suffix tree combine filter factor[J]. Computer Applications and Software, 2022, 39(1): 266-270(in Chinese).
[21]	DEVLIN J, CHANG M W, LEE K, et al. BERT: Pre-training of deep bidirectional transformers for language understanding[EB/OL]. (2006-01-24)[2022-07-03]2018: arXiv: 1810.04805.
[22]	张华, 叶娜, 周俏丽, 等. 基于分类策略的术语识别系统融合[J]. 小型微型计算机系统, 2015, 36(2): 385-390. ZHANG H, YE N, ZHOU Q L, et al. Classification strategy based term recognition systems combination[J]. Journal of Chinese Computer Systems, 2015, 36(2): 385-390(in Chinese).
[23]	LEVY R, MANNING C. Is it harder to parse Chinese, or the Chinese Treebank? [C]//Proceedings of the 41st Annual Meeting on Association for Computational Linguistics-ACL '03. Morristown: Association for Computational Linguistics, 2003: 439-446.
[24]	PASZKE A, GROSS S, MASSA F, et al. PyTorch: An imperative style, high-performance deep learning library[EB/OL]. (2022-07-03)[2019-12-03].
[25]	黄勋, 游宏梁, 于洋. 关系抽取技术研究综述[J]. 现代图书情报技术, 2013(11): 30-39. HUANG X, YOU H L, YU Y. A review of relation extraction[J]. New Technology of Library and Information Service, 2013(11): 30-39(in Chinese).