Automatic classification of civil aviation safety information based on text augment

Abstract

Abstract: Aiming at the problem of insufficient samples in the application of automatic classification of civil aviation safety information, an automatic classifier is developed, based on the bidirectional encoder representations from transformers (BERT) pre-training model, easy data augment(EDA) and support vector machine(SVM) algorithms.This paper categorizes the incident information into data subsets of different orders of magnitude according to the orders of magnitude of information of single incident type and analyzes the model performance of data-sets in different scale, especially for the small data-sets. The results show that the F 1w value is increased by 31.21 % , 9.66 % and 3.35 % when the orders of magnitude of single incident type are ten -scale, hundred -scale and thousand -scale respectively, indicating a significant improvement of model performance. Therefore, by text enhancement algorithm, the automatic classifier trained in relatively small data-sets has a good effect, and is applicable to automatic classification and processing for civil aviation safety information.

Key words: civil aviation safety, safety information, text augment, natural language processing

CLC Number:

V328
X949

CUI Zhenxin, ZHANG Zhuoyan. Automatic classification of civil aviation safety information based on text augment[J]. Journal of Civil Aviation University of China, 2022, 40(3): 47-53.

References 12

[1]	史亚杰, 陈艳秋. 航空安全信息管理的问题与对策[J]. 中国安全生产科学技术, 2010, 6(3): 116-120.
[2]	ZHANG X, ZHAO J B, LECUN Y. Character-level convolutional networks for text classification[C]//28th International Conference on Neural Information Processing Systems, September 4, 2015, Montreal, Quebec,Canada. Cambridge: MIT Press, 2015: 649-657.
[3]	KOBAYASHI S. Contextual augmentation: data augmentation by words with paradigmatic relations[C]//2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers),June 1-6, 2018, New Orleans, Louisiana. Stroudsburg: Association for Computational Linguistics, 2018.
[4]	YU A W, DOHAN D, LUONG M-T, et al QANet: combining local convolution with global self -attention for reading comprehension [C]//6th International Conference on Learning Representations Vancouver Convention Center, Vancouver, BC, Canada, April 30 - May 3, 2018.
[5]	MIKOLOV T, CHEN K, CORRADO G, et al Efficient estimation of word representations in vector space[C]//International Conference on Learning Representations, Scottsdale, Arizona, USA, May 2-4, 2013.
[6]	PETERS M E, NEUMANN M, IYYER M, et al Deep contextualized word representations[EB/OL]. (2018-05-23)[2020-07-02]. https://arxiv.org/pdf/1802.05365.pdf.
[7]	RADFORD A, NARASIMHAN K, SALIMANS T, et al Improving language understanding by generative pre-training[EB/OL]. [2020-07-20]. https://www.cs.ubc.ca/~amuham01/LING530/papers/radford2018improving.pdf.
[8]	DEVLIN J, CHANG M W, LEE K, et al BERT: Pre-training of deep bidirectional transformers for language understanding[C]//2019 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Minneapolis, MN, June 2-7, 2019: 4171-4186.
[9]	JOACHIMS T. Text categorization with Support Vector Machines: learning with many relevant features[C]//10th European Conference on Machine Learning, Chemnitz, Germany, April 21-23, 1998: 137-142.
[10]	WEI J, ZOU K． EDA: easy data augmentation techniques for boosting performance on text classification tasks[C]//2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), Hong Kong, China． Stroudsburg, PA, USA: Association for Computational Linguistics, 2019: 6382-6388．
[11]	ZHAN L L. EDA_NLP_for_Chinese[EB/OL]. (2019-05-17) [2020-07-05]. https://github.com/zhanlaoban/eda_nlp_for_Chinese.
[12]	TURC I, CHANG M W, LEE K, et al Well-read students learn better: on the importance of pre-training compact models[EB/OL]. (2019-09-25)[2020-03-09]. https://arxiv.org/abs/1908.08962.