Journal of Engineering, Project, and Production Management, 2026, 16(4), 2025-219
Optimization and Visualization of Industrial Production Knowledge Graph Based on BERT and RGCN
Associate Professor, School of Law and School of Intellectual Property, Tiangong University, Tianjin, 300387, China, E-mail: tjgydxresearch2004@163.com
Project Management
Received October 10, 2025; revised December 22, 2025; accepted December 23, 2025
Available online May 29, 2026
Abstract: To improve the accuracy and efficiency of existing methods for optimizing and visualizing multivariate data knowledge graphs, a new model is proposed. The industrial production knowledge graph optimization and visualization model integrates a Bidirectional Encoder Representation (BERT) model, a Relational Graph Convolutional Network (RGCN), and Scalable Vector Graphics (SVG) technology. This model aims to enhance the accuracy of data fusion and semantic relationship mining by optimizing the bidirectional encoder representation through Conditional Random Fields (CRF). It also optimizes the knowledge graph using relational graph convolutional networks and Graph Attention Networks (GAN), and combines Scalable Vector Graphics (SVG) technology to complete its visualization. The main innovation of the research lies in overcoming the limitations of the existing method’s insufficient adaptability to diverse industrial data by constructing a collaborative framework that features semantic enhancement, knowledge graph optimization, and efficient visualization. The study compared the proposed model with the optimization model based on deep learning. The results show that the proposed model’s recall rate for equipment fault identification after optimizing the knowledge graph is 98.22%. The resource utilization rate and response delay during optimization are 27.81% and 12.8 minutes, respectively. Meanwhile, the average efficiency of this model in visualizing multiple-entity data is 96.69%, 93.66%, and 96.26%, respectively, with the structural clarity of visualizing the temperature data in the operation log at 97.26%. All experimental results outperform the comparison models, fully proving the feasibility and advantages of the proposed model. This study provides new ideas and methods for the further development of knowledge graphs.
Keywords: Multisource data, CRF-BERT, semantic relations, GAT-RGCN, knowledge graph visualization, scalable vector graphics.
Copyright © Journal of Engineering, Project, and Production Management (EPPM-Journal).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Requests for reprints and permissions at eppm.journal@gmail.com.
Citation: Li, S. (2026). Optimization and Visualization of Industrial Production Knowledge Graph Based on BERT and RGCN. Journal of Engineering, Project, and Production Management, 16(4), 2025-219.
