Article: Keyword association mining and topic evolution analysis algorithm for knowledge graph in the energy field Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.69 pp.45 - 68 Abstract: With the rapid growth of energy literature data, accurately mining semantic associations between keywords and dynamically tracking topic evolution patterns is a key challenge. This paper proposes an algorithm for keyword association mining and topic evolution analysis for knowledge graphs in the energy field. The model integrates topic modelling, graph neural networks and time series analysis. It combines the topic probability distribution from BERTopic with the knowledge graph topology via a topic- graph coupling mechanism, uses a graph attention network to optimise association weights. Test results show the model outperforms baseline models like LDA and BERTopic in accuracy (91.2%), F1-score (0.892) and topic consistency (0.848). It also excels in robustness (F1-score drops 7.2% with 20% noise), interpretability (expert score 4.5/5) and generalisation (performance degradation 6.3%). These results verify the model's efficiency and reliability for practical energy knowledge analysis, providing support for energy policy evaluation and technology trend prediction. Inderscience Publishers - linking academia, business and industry through research

Title: Keyword association mining and topic evolution analysis algorithm for knowledge graph in the energy field

Authors: Wenke Li; Xinping Miao; Qingbo Kong; Yue Tian; Ruoyan Dong

Addresses: Guizhou Power Grid Co., Ltd., Digital Intelligence Operation Center, Gui Yang Gui Zhou, 551400, China ' Guizhou Power Grid Co., Ltd., Digital Intelligence Operation Center, Gui Yang Gui Zhou, 551400, China ' Guizhou Power Grid Co., Ltd., Digital Intelligence Operation Center, Gui Yang Gui Zhou, 551400, China ' Guizhou Power Grid Co., Ltd., Digital Intelligence Operation Center, Gui Yang Gui Zhou, 551400, China ' Guizhou Power Grid Co., Ltd., Digital Intelligence Operation Center, Gui Yang Gui Zhou, 551400, China

Abstract: With the rapid growth of energy literature data, accurately mining semantic associations between keywords and dynamically tracking topic evolution patterns is a key challenge. This paper proposes an algorithm for keyword association mining and topic evolution analysis for knowledge graphs in the energy field. The model integrates topic modelling, graph neural networks and time series analysis. It combines the topic probability distribution from BERTopic with the knowledge graph topology via a topic- graph coupling mechanism, uses a graph attention network to optimise association weights. Test results show the model outperforms baseline models like LDA and BERTopic in accuracy (91.2%), F1-score (0.892) and topic consistency (0.848). It also excels in robustness (F1-score drops 7.2% with 20% noise), interpretability (expert score 4.5/5) and generalisation (performance degradation 6.3%). These results verify the model's efficiency and reliability for practical energy knowledge analysis, providing support for energy policy evaluation and technology trend prediction.

Keywords: knowledge graph of energy field; keyword association mining; theme evolution analysis; graph neural network; GNN; dynamic topic model; DTM.

DOI: 10.1504/IJICT.2026.154381

International Journal of Information and Communication Technology, 2026 Vol.27 No.69, pp.45 - 68

Received: 30 Jan 2026
Accepted: 17 Mar 2026
Published online: 25 Jun 2026 *

Title: Keyword association mining and topic evolution analysis algorithm for knowledge graph in the energy field

Keep up-to-date