Exploration of Enhancement Effect in Natural Language Understanding Task Based on BERT Model with Integrated Power Knowledge Graph

Authors

  • Weidong Guo State Grid Shanxi Electric Power Company, Taiyuan 030001, Shanxi, China
  • Yubin Xu State Grid Shanxi Electric Power Company, Taiyuan 030001, Shanxi, China
  • Zhifeng Shi State Grid Shanxi Electric Power Company, Taiyuan 030001, Shanxi, China
  • Xuefeng Hu State Grid Shanxi Electric Power Company, Taiyuan 030001, Shanxi, China
  • Huiping Zhang State Grid Shanxi Electric Power Company, Taiyuan 030001, Shanxi, China
  • Lei Feng State Grid Shanxi Electric Power Company Material Branch, Taiyuan 030001, Shanxi, China

Abstract

In response to the problems of poor domain adaptability and the weak knowledge fusion ability of traditional language models in NLU (Natural Language Understanding) tasks, which prevent them from fully capturing deep relationships in context, this studyaims to integrate power knowledge graphs to enhance the effectiveness of BERT (Bidirectional Encoder Representations from Transformers) models in NLU tasks, enabling them to more accurately infer the terminology and contextual meanings pertaining tothe electricityfield, thereby improving training efficiency and model performance, and promoting the development of automation in this field. The performance evaluation results of the BERT model integrating power knowledge graph in NLU tasks were: an average path length of 3.8, language similarity of 0.9, and vocabulary coverage of 0.8, all of which were superior to other models used for comparison. The experimental results showed that the BERT model integrating a power knowledge graph had better performance compared to other models commonly used for processing NLU tasks.

Keywords: power knowledge graph; BERT model; natural language understanding; power sector; knowledge fusion

Cite As

W. Guo, Y. Xu, Z. Shi, X. Hu, H. Zhang, L. Feng, "Exploration of Enhancement Effect in Natural Language Understanding Task
Based on BERT Model with Integrated Power Knowledge Graph", Engineering Intelligent Systems, vol. 34 no. 1, pp. 17-26, 2026.



 

Downloads

Published

2026-01-01

Issue

Vol. 34 No. 1 (2026): International Journal of Engineering Intelligent Systems

Section

Article

License

The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.