Research on the Construction and Application of Knowledge Graph in Materials Science

doi:10.11871/jfdc.issn.2096-742X.2025.01.011

Abstract

Abstract:

[Objective] To facilitate the processing of large-scale unstructured materials science information and to deeply explore the relationship between materials information data, knowledge mapping provides technical support for the integrated study of materials' properties and structures. [Methods] In this study, a new knowledge graph construction method is proposed, and a knowledge graph-based material knowledge intelligent question and answer system and a material knowledge graph cloud platform are constructed. The knowledge intelligent question and answer system is built using BERT+CRF, and the graph cloud platform is developed based on the B/S framework. [Results] The system framework constructed in this study can realize the knowledge question and answer of materials, provide users with the entrance to the cloud platform, and visualize the material knowledge graph. Taking the III-V compound computational dataset as an example, we constructed the III-V compound material knowledge graph. [Conclusions] This study extends the application of knowledge graph in the field of materials, which is important for accelerating the discovery of new materials and mining potential information about materials.

Key words: knowledge graph, materials data, intelligent question-and-answer, cloud platform

YUAN Yang, LIU Qilin, CHEN Ziyi, WAN Meng, LI Kai, WANG Yangang, WANG Jing, WANG Zongguo. Research on the Construction and Application of Knowledge Graph in Materials Science[J]. Frontiers of Data and Computing, 2025, 7(1): 152-162, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2025.01.011.

Figures/Tables 12

Table 1

Fig.1

Table 2

Fig.2

Fig.3

Table 3

Table 4

Fig.4

Table 5

Fig.5

Fig.6

Fig.7

References 15

[1]	ZHANG X, LIU X, LI X, et al. MMKG: an approach to generate metallic materials knowledge graph based on DBpedia and Wikipedia[J]. Computer Physics Communications, 2017, 211: 98-112.
[2]	VENUGOPAL V, OLIVETTI E. MatKG: An autonomously generated knowledge graph in Material Science[J]. Scientific Data, 2024, 11(1): 217.
[3]	JAIN A, ONG S P, HAUTIER G, et al. Commentary: The Materials Project: A materials genome approach to accelerating materials innovation[J]. APL Materials, 2013, 1(1): 011002.
[4]	NIE Z, LIU Y, YANG L, et al. Construction and application of materials knowledge graph based on author disambiguation: revisiting the evolution of LiFePO4[J]. Advanced Energy Materials, 2021, 11(16): 2003580.
[5]	NIE Z, ZHENG S, LIU Y, et al. Automating materials exploration with a semantic knowledge graph for Li-ion battery cathodes[J]. Advanced Functional Materials, 2022, 32(26): 2201437.
[6]	ZHAO X, GREENBERG J, MCCLELLAN S, et al. Knowledge graph-empowered materials discovery[C]// 2021 IEEE International Conference on Big Data (Big Data). IEEE, 2021: 4628-4632.
[7]	BELSKY A, HELLENBRANDT M, KAREN V L, et al. New developments in the Inorganic Crystal Structure Database (ICSD): accessibility in support of materials research and design[J]. Acta Crystallographica Section B: Structural Science, 2002, 58(Pt 3 Pt 1): 364-369.
[8]	DEVLIN J, CHANG M W, LEE K, et al. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding[C]// Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies, Volume 1 (Long and Short Papers). Minneapolis: Association for Computational Linguistics, 2019: 4171-4186.
[9]	MA X, HOVY E. End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1:Long Papers). Berlin:Association for Computational Linguistics, 2016: 1064-1074.
[10]	LÓPEZ F M S, DE LA CRUZ E G S. Literature review about Neo4j graph database as a feasible alternative for replacing RDBMS[J]. Industrial Data, 2015, 18(2): 135-139.
[11]	KIRKLIN S, SAAL J E, MEREDIG B, et al. The Open Quantum Materials Database (OQMD): assessing the accuracy of DFT formation energies[J]. npj Computational Materials, 2015, 1(1): 1-15.
[12]	SCHMIDHUBER J, HOCHREITER S. Long short-term memory[J]. Neural Computing, 1997, 9(8): 1735-1780.
[13]	MIKOLOV T, SUTSKEVER I, CHEN K, et al. Distributed representations of words and phrases and their compositionality[J]. Advances in Neural Information Processing Systems, 2013, 26.
[14]	Cai X, Sun E, Lei J. Research on application of named entity recognition of electronic medical records based on BERT-IDCNN-CRF model[C]// Proceedings of the 6th International Conference on Graphics and Signal Processing, 2022: 80-85.
[15]	SANG T K, ERIK F. Introduction to the CoNLL-2003 shared task: Language-independent named entity recognition[C]// Proceedings of the Seventh Conference on Natural Language Learning at HLT-NAACL 2003, 2003: 142-147.

头节点	节点关系	属性节点
Hash_Value	has_Gap	能隙
data_id	has_EnergyPerAtom	原子能
	has_Formula	分子式
	has_NumberOfSites	点位数
	has_Crystalsystem	晶系
	has_NumberOfElements	元素种类个数
	has_Volume	体积

数据条数	节点数	关系数	串行开销（s）	并行开销（s）
5,000	22,177	62,538	175.5	62.6
10,000	38,746	114,079	339.1	126.2
15,000	60,965	181,307	558.3	190.7
20,000	77,553	236,667	775.4	253.7
25,000	97,252	305,769	1,015.6	317.9
30,000	117,405	375,003	1,280.9	385.8
33,887	129,381	418,423	1,612.2	449.1

模版类型	语料示例
陈述句	氧化铝的分子式为Al₂O₃
陈述句	LiAlAs₂O₇的体积为124.25 ?³
问句	形成能大于-3小于-2的材料有哪些
问句	分子式为Bi₄B₁₂O₂₄的形成能是多少

模型	指标	精确率	召回率	F1分数
BERT	macro avg	0.9429	0.9435	0.9423
	weighted avg	0.9818	0.9824	0.9818
BERT+CRF	macro avg	0.9845	0.9834	0.9838
	weighted avg	0.9844	0.9845	0.9844