Article Details

Abstract

Taking AI patents as the research object, this research integrates a multi-dimensional indicator system with Large Language Models to achieve the objectives of topic mining and evolutionary trend identification. First, multi-dimensional indicators are constructed to screen and analyze core patent clusters. Second, LLMs are utilized for dynamic topic mining of the core patent clusters. The results indicate that technological topics in the field of artificial intelligence exhibit an evolutionary pattern characterized by “emergence, expansion, explosion, integration, and stabilization”, with topics demonstrating concentrated interaction within disciplines and active diffusion across disciplines. This research primarily addresses issues related to data processing time and algorithm compatibility, while limitations remain in data source diversity and model threshold settings, necessitating further in-depth research.

Keywords

multidimensional indicator; Large Language Model; Artificial Intelligence; topic mining; evolution trend

References

1. [1] Zhao Qizhi, Mo Zan Research on Emerging Technology Identification Methods Based on Multi source Data and Multi dimensional Indicators: Taking New Energy Vehicle Technology as an Example [J]. Journal of Guangxi University (Philosophy and Social Sciences Edition), 2025, 47 (2): 187-196
2. [2] Yang Hangyu, Wu Zengyuan, Han Xiangli, etc Research on Key Technology Identification of "Stuck Neck" Based on Patent Mining: Taking the Field of Analytical Instruments as an Example [J]. Science and Technology Management Research, 2025, 45 (4): 30-38
3. [3] Li Yuhua, Zhang Fujun, Yin Yanxia, etc Quantitative Analysis of Patent Information in the Global Artificial Intelligence Science Field and Its Implications [J]. Science and Technology Management Research, 2020, 40 (21): 154-162
4. [4] Zhang Biao, Chen Yunwei, Dong Kun Research on Enterprise Technology Opportunity Discovery Method Based on Link Prediction and Multidimensional Indicators [J]. Data Analysis and Knowledge Discovery, 2024, 8 (7): 137-148
5. [5] Liu Z F，Wang X Y，Feng J. Generating Firm-centric Technology Opportunities from Patent Recommendation: A Dual-link Prediction Perspective [J].Technology Analysis and Strategic Management，2026，38（7）：925-942.
6. [6] Wu Ji, Xu Pingping, Yang Xiao, etc Evaluation and Analysis of the Development of Quantum Computing Frontier Technologies Based on Scientific Metrology [J]. Science and Technology Management Research, 2025, 45 (19): 9-21
7. [7] Lu Wanhui Research on the Characteristics of Science Intensive Technology Research and Development Based on Patent Metrology [J]. Intelligence Theory and Practice, 2023, 46 (2): 40-48
8. [8] Sun Minghan, Zhu Xiuzhu Research on the main path of fuel cell technology based on the same family patent citation network [J]. Intelligence Journal, 2022, 41 (12): 51-56
9. [9] Liu Xiang, Wan Xiaoping, Yan Xiaoting, etc Main path analysis based on citation path overlay network [J]. Journal of Intelligence, 2019, 38 (8): 807-814
10. [10] Feng Lijie, Zhou Wei, Liu Peng, etc Research on Technology Evolution Path Identification and Expansion Based on Citation Network and Semantic Analysis [J]. Intelligence Theory and Practice, 2023, 46 (1): 90-99131
11. [11] Xiong Yan, Li Xiangyu, Hu Kai Research on the Dynamic Evolution of Global Chip Industry Innovation Trends and Cooperation Networks: Taking Logic Chips as an Example [J]. China Science and Technology Forum, 2025 (12): 177-188
12. [12] Wei Qifeng, Yang Ding Research on the Characteristics and Evolution of China's Chip Technology Innovation Cooperation Network [J]. Research Management, 2026, 47 (7): 45-55
13. [13] Zhao R B，Wei W Y，Zhu Z J. Discovering firm Technology Opportunities by Machine Learning Prediction Method Based on Firm's Technology Portfolio [J/OL].Technological Forecasting and Social Change，2025，221：124305.https://www.sciencedirect.com/science/article/abs/pii/S0040162525003361?via%3Dihub.
14. [14] Deng Na, but Wen Jun, Chen Xu, etc EDGCL: A Chinese medicine patent technology link prediction method based on explicit implicit bisexuality association and graph contrastive learning [J/OL]. Data Analysis and Knowledge Discovery, 2026:1-20 https://link.cnki.net/urlid/10.1478.G2.20251117.1747.008.
15. [15] Yang Liangliang, Yao Shuang, Li Yueyan, etc Research on Technology Opportunity Identification of Link Prediction and Multidimensional Evaluation Based on GCN Two stage Feature Construction [J]. Data Analysis and Knowledge Discovery, 2025, 9 (12): 1-17
16. [16] Zhang Yanlu, Ding Zeyu, Yang Naiding, etc Research on the identification method of key core technologies for "bottleneck" based on a two-stage screening model: taking the aviation manufacturing industry as an example [J]. Science and Technology Management Research, 2025, 45 (20): 149-158
17. [17] Ran Congjing, Tian Wenfang Research on Potential Emerging Technologies Identification by Integrating SVM-LDA and Weighted Similarity: A Case Study in the Field of Artificial Intelligence [J]. Journal of Intelligence, 2024, 43 (5): 563-574
18. [18] Liu Jianhua, Chen Zihuan, Wang Huiyang Key core technology identification integrating dual layer patent mining and technical structure - taking superhard materials as an example [J]. Intelligence Journal, 2025, 44 (12): 107-117
19. [19] Luan G，Yang G，Zhen S，et al. Emerging Disruptive Technologies in Tourism: An LLM-based Identification Framework [J/OL].Current Issues in Tourism，2025（12）：1-20.https://doi.org/10.1080/13683500.2025.2581155.
20. [20] Li Jinfa, Xu Can, Zhang Ke, etc Research on Emerging Technology Recognition Integrating Deep Learning and Patent Maps in Dynamic Perspective [J]. Intelligence Theory and Practice, 2025, 48 (7): 160-169
21. Zhao Yiming, Liu Shunsheng, Lv Lucheng Research on Early Identification of Disruptive Technologies Based on Ensemble Learning: A Case Study of Quantum Computing [J]. Data Analysis and Knowledge Discovery, 2025, 9 (10): 85-98
22. [22] Wang Haijun, Yu Jiawen Research on Disruptive Technology Theme Recognition Based on Patents and Weibo: Taking the Field of Artificial Intelligence as an Example [J]. China Science and Technology Forum, 2024 (7): 83-94109
23. [23] China National Intellectual Property Administration Classification System for Key Digital Technology Patents (2023) [EB/OL]. (2023-08-17) [2025-06-25] https://www.cnipa.gov.cn/attach/0/d32119ae1faa4fbf9e308824862f87dd.pdf.

Funding

General Project of Philosophy and Social Sciences Planning in Guangdong Province (GD24CGL57); Special Project of Qingyuan Science and Technology Think Tank "Analysis and Optimization of Qingyuan New Materials Industry Technology Competitiveness from the Perspective of Neural Symbol Integration Knowledge Graph Reasoning" (QYZKYB-2025-02)

About the authors

Wu Qingyuan (1985-), female, Han ethnicity, from Loudi, Hunan, lecturer, doctoral student, research interests include scientific and technological innovation and data mining.