With the rapid development in AI technology, constructing the nationally independent intelligent railway is a trend at present in Chinese rail transit industry. As CTCS-3 is one of the core technologies for Chinese high-speed railway, making machines understand the CTCS-3 knowledge efficiently and concretely is becoming an important topic. Knowledge extraction is one of the most significant parts. Therefore, we proposed a method to extract CTCS-3 knowledge from unstructured data by combining BERT and BiLSTM-CRF. We built a 407936-word labeled dataset about CTCS-3 equipment for model training. Using such small dataset, we completed the experiments of CTCS-3 knowledge extraction including entity recognition and relationship extraction. The experiment show that the F1 score is 75.89%. By the method, we can get some entity relationship triples which are the foundation to achieve cognitive intelligence of CTCS-3. In summary, it extracts CTCS-3 entity relation triples with a small number of rail transit industry dataset.
A Method for CTCS-3 Knowledge Extraction of Unstructured Data
Lect. Notes Electrical Eng.
International Conference on Electrical and Information Technologies for Rail Transportation ; 2021 October 21, 2021 - October 23, 2021
Proceedings of the 5th International Conference on Electrical Engineering and Information Technologies for Rail Transportation (EITRT) 2021 ; Chapter : 8 ; 67-74
2022-02-19
8 pages
Article/Chapter (Book)
Electronic Resource
English
A Method for CTCS-3 Knowledge Extraction of Unstructured Data
British Library Conference Proceedings | 2022
|Method for automatically switching train-mounted equipment between CTCS-0 level and CTCS-4 level
European Patent Office | 2021
|CTCS-Chinese Train Control System
British Library Conference Proceedings | 2004
|Middleware-Based Distributed CTCS-3 Simulation Platform
Tema Archive | 2014
|