Extraction and analysis of risk factors from Chinese chemical accident reports

doi:10.1016/j.cjche.2023.02.026

Abstract

Abstract: Accidents in chemical production usually result in fatal injury, economic loss and negative social impact. Chemical accident reports which record past accident information, contain a large amount of expert knowledge. However, manually finding out the key factors causing accidents needs reading and analyzing of numerous accident reports, which is time-consuming and labor intensive. Herein, in this paper, a semi-automatic method based on natural language process (NLP) technology is developed to construct a knowledge graph of chemical accidents. Firstly, we build a named entity recognition (NER) model using SoftLexicon (simplify the usage of lexicon) + BERT-Transformer-CRF (conditional random field) to automatically extract the accident information and risk factors. The risk factors leading to accident in chemical accident reports are divided into five categories: human, machine, material, management, and environment. Through analysis of the extraction results of different chemical industries and different accident types, corresponding accident prevention suggestions are given. Secondly, based on the definition of classes and hierarchies of information in chemical accident reports, the seven-step method developed at Stanford University is used to construct the ontology-based chemical accident knowledge description model. Finally, the ontology knowledge description model is imported into the graph database Neo4j, and the knowledge graph is constructed to realize the structured storage of chemical accident knowledge. In the case of information extraction from 290 Chinese chemical accident reports, SoftLexicon + BERT-Transformer-CRF shows the best extraction performance among nine experimental models. Demonstrating that the method developed in the current work can be a promising tool in obtaining the factors causing accidents, which contributes to intelligent accident analysis and auxiliary accident prevention.

Key words: Chemical processes, Chemical process safety, Natural language process, Knowledge graph, Neural networks, Algorithm

摘要： Accidents in chemical production usually result in fatal injury, economic loss and negative social impact. Chemical accident reports which record past accident information, contain a large amount of expert knowledge. However, manually finding out the key factors causing accidents needs reading and analyzing of numerous accident reports, which is time-consuming and labor intensive. Herein, in this paper, a semi-automatic method based on natural language process (NLP) technology is developed to construct a knowledge graph of chemical accidents. Firstly, we build a named entity recognition (NER) model using SoftLexicon (simplify the usage of lexicon) + BERT-Transformer-CRF (conditional random field) to automatically extract the accident information and risk factors. The risk factors leading to accident in chemical accident reports are divided into five categories: human, machine, material, management, and environment. Through analysis of the extraction results of different chemical industries and different accident types, corresponding accident prevention suggestions are given. Secondly, based on the definition of classes and hierarchies of information in chemical accident reports, the seven-step method developed at Stanford University is used to construct the ontology-based chemical accident knowledge description model. Finally, the ontology knowledge description model is imported into the graph database Neo4j, and the knowledge graph is constructed to realize the structured storage of chemical accident knowledge. In the case of information extraction from 290 Chinese chemical accident reports, SoftLexicon + BERT-Transformer-CRF shows the best extraction performance among nine experimental models. Demonstrating that the method developed in the current work can be a promising tool in obtaining the factors causing accidents, which contributes to intelligent accident analysis and auxiliary accident prevention.

关键词: Chemical processes, Chemical process safety, Natural language process, Knowledge graph, Neural networks, Algorithm

Xi Luo, Xiayuan Feng, Xu Ji, Yagu Dang, Li Zhou, Kexin Bi, Yiyang Dai. Extraction and analysis of risk factors from Chinese chemical accident reports[J]. Chinese Journal of Chemical Engineering, 2023, 61(9): 68-81.

Xi Luo, Xiayuan Feng, Xu Ji, Yagu Dang, Li Zhou, Kexin Bi, Yiyang Dai. Extraction and analysis of risk factors from Chinese chemical accident reports[J]. 中国化学工程学报, 2023, 61(9): 68-81.

References

[1] Somayeh, Mirzaei, A comparative study on long short-term memory and gated recurrent unit neural networks in fault diagnosis for chemical processes using visualization, J. Taiwan Inst. Chem. Eng. 130 (2022) 104028.
[2] Qiusheng, Song, Jiang, A multi-scale convolutional neural network based fault diagnosis model for complex chemical processes, Process. Saf. Environ. Prot. 159 (2022) 575–584.
[3] Wu, Hao, Deep convolutional neural network model based chemical process fault diagnosis, Comput. Chem. Eng. 115 (2018) 185–197.
[4] L Ming, J.S. Zhao, Feature selection for chemical process fault diagnosis by artificial immune systems, Chin. J. Chem. Eng. 26 (8) (2018) 1599–1604.
[5] Xiaoyong, Gao, A decision tree based decomposition method for oil refinery scheduling, Chin. J. Chem. Eng. 26 (8) (2018) 1605–1612.
[6] W.D. Tian, G.X. Zhang, X. Zhang, Y.X. Dong, PCA weight and Johnson transformation based alarm threshold optimization in chemical processes, Chin. J. Chem. Eng. 26 (2018) (8)1653–1661.
[7] M.Y. Li, W.L. Du, F. Qian, W.M. Zhong, Total plant performance evaluation based on big data: Visualization analysis of TE process, Chin. J. Chem. Eng. 26 (2018) (8)1736–1749.
[8] K.K Peng, Q.K. Pan, B Zhang, An improved artificial bee colony algorithm for steelmaking-refining-continuous casting scheduling problem, Chin. J. Chem. Eng. 26 (8) (2018) 1727–1735.
[9] X.L. Wang, L Huang, C.H. Yang, Prediction model of slurry pH based on mechanism and error compensation for mineral flotation process, Chin. J. Chem. Eng. 26 (8) (2018) 1766–1772.
[10] L. Jia, W. Tan, Just-in-time learning based integrated MPC-ILC control for batch processes, Chin. J. Chem. Eng. 26 (2018) (8)1713–1720.
[11] Y.L. Ge, S.R. Li, X.D. Zhang, Optimization for ASP flooding based on adaptive rationalized Haar function approximation, Chin. J. Chem. Eng. 26 (2018) (8)1758–1765.
[12] I, Chalkidis, M. Fergadiotis, P. Malakasiotis, N. Aletras, I. Androutsopoulos, LEGAL-BERT: The Muppets straight out of law school. Findings of the Association for Computational Linguistics: EMNLP 2020 (2020) 2898–2904.
[13] J. Lee, W. Yoon, S. Kim, D. Kim, S. Kim, C.H. So, J. Kang, BioBERT: a pre-trained biomedical language representation model for biomedical text mining, Bioinformatics 36 (4) (2020) 1234–1240.
[14] Na, XU, An improved text mining approach to extract safety risk factors from construction accident reports, Saf. Sci. 138 (2021) 105216.
[15] L.L. Hua, W. Zheng, S.G. Gao, Extraction and analysis of risk factors from Chinese railway accident reports, 2019 IEEE Intelligent Transportation Systems Conference (ITSC). Auckland, New Zealand. IEEE, (2019) 869–874.
[16] J. Devlin, M.W. Chang, K. Lee, K. Toutanova, BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Minneapolis, Minnesota, 2019.
[17] A.J.-P, Tixier, Automated content analysis for construction safety: a natural language processing system to extract precursors and outcomes from unstructured injury reports, Autom. Constr. 62 (2016) 45–56.
[18] X.Y. Feng, Y.Y. Dai, X. Ji, L. Zhou, Y.G.Dang, Application of natural language processing in HAZOP reports, Process. Saf. Environ. Prot. 155 (2021) 41–48.
[19] G.Y. Liu, M. Boyd, M. Yu, S. Z. Halim, N. Quddus, Identifying causality and contributory factors of pipeline incidents by employing natural language processing and text mining techniques, Process. Saf. Environ. Prot. 152 (2021) 37–46.
[20] Fan, Zhang, Construction site accident analysis using text mining and natural language processing techniques, Autom. Constr. 99 (2019) 238–248.
[21] W. Fang, Automated text classification of near-misses from safety reports: an improved deep learning approach, Adv. Eng. Inform. 44 (2020) 101060.
[22] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A.N. Gomez, L. Kaiser, I. Polosukhin, Attention is all you need, In: Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS17), Long Beach, CA, USA, 2017.
[23] J.I. Single, J Schmidt, J Denecke, Knowledge acquisition from chemical accident databases using an ontology-based method and natural language processing, Saf. Sci. 129 (2020) 104747.
[24] B.S. Song, F. Li, Y.S. Liu, X.X. Zeng, Deep learning methods for biomedical named entity recognition: a survey and qualitative comparison, Brief Bioinform 22 (6) (2021) bbab282.
[25] L.S. Li, R.P. Zhou, D.G.Huang, Two-phase biomedical named entity recognition using CRFs, Comput. Biol. Chem. 33 (4) (2009) 334–338.
[26] J.D. Lafferty, A. McCallum, F.C.N. Pereira, Conditional random fields: probabilistic models for segmenting and labeling sequence data, Proceedings of the Eighteenth International Conference on Machine Learning. New York: ACM, (2001) 282–289.
[27] L. Luo, Z.H. Yang, P. Yang, Y. Zhang, L. Wang, H.F. Lin, J. Wang, An attention-based BiLSTM-CRF approach to document-level chemical named entity recognition, Bioinformatics 34 (8) (2018) 1381–1388.
[28] Dan, Feng, A small samples training framework for deep Learning-based automatic information extraction: case study of construction accident news reports analysis, Adv. Eng. Inform. 47 (2021) 101256.
[29] Bingyu, Wang, Automatic frequency estimation of contributory factors for confined space accidents, Process. Saf. Environ. Prot. 157 (2022) 193–207.
[30] Shuai, Mao, Development of process safety knowledge graph: a Case study on delayed coking process, Comput. Chem. Eng. 143 (2020) 107094.
[31] Z. Huang, W. Xu, K. Yu, Bidirectional LSTM-CRF models for sequence tagging, arXiv preprint arXiv: 1508.01991, 2015.
[32] Y. Zhang, J.Yang, Chinese NER Using Lattice LSTM.Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Melbourne, Australia. Stroudsburg, PA, USA: Association for Computational Linguistics, 2018.
[33] R.T. Ma, M.L. Peng, Q. Zhang, Z.Y. Wei, X.J.Huang, Simplify the Usage of Lexicon in Chinese NER.Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Online. Stroudsburg, PA, USA: Association for Computational Linguistics, 2020.
[34] Y.M. Cui, W.X. Che, T. Liu, B. Qin, Z.Q. Yang, Pre-training with whole word masking for Chinese BERT, IEEE/ACM Trans. Audio Speech Lang. Process. 29 (2021) 3504–3514.