Technological innovation serves as a powerful driving force for the high-quality development of industries. This study proposes a method for identifying cutting-edge technologies based on AI agent and multi-feature text fusion, offering insights into grasping directions for technological breakthroughs and exploring opportunities for technological innovation. Using funding projects, academic papers, and patent data as research objects, this approach first applies AI agent for data preprocessing, extracting textual keywords. Then, the semantic features, classification features, and Biterm topic features of scientific and technological documents are fused to represent the text vector and enhance the data features. The clustering algorithm is used to divide the technical topics. Finally, a cutting-edge technology measurement index system based on novelty, growth, influence, and innovation is constructed to identify cuttingedge technologies. An empirical study is conducted in the field of micro-LED. Clustering methods that integrate semantic, classification, and topic features demonstrate superior performance compared to using only single or dual features. The analysis identifies emerging frontier technologies such as chip manufacturing technology and full-color display technology; highlights popular frontier technologies including light-emitting materials and devices, display technology, holographic display, medical applications, detection and repair, and driving circuits; and also uncovers potential frontier technologies like mass transfer technology. Thus, the effectiveness and application value of the proposed method are proved.
关键词
前沿技术识别/AI Agent/多特征融合/Biterm主题模型
Key words
Cutting-Edge Technology Identification/AI Agent/Multi-Feature Fusion/Biterm Topic Model