数字图书馆论坛2026，Vol.22Issue(1)：31-43,13.DOI:10.3772/j.issn.1673-2286.2026.01.004
基于AI Agent和文本多特征融合的前沿技术识别方法研究
Cutting-Edge Technology Identification Method Based on AI Agent and Multi-Feature Text Fusion
摘要
Abstract
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 cutting-edge 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分类
社会科学引用本文复制引用
李贞贞,胡思思,钟永恒,王辉,刘佳..基于AI Agent和文本多特征融合的前沿技术识别方法研究[J].数字图书馆论坛,2026,22(1):31-43,13.基金项目
本研究得到湖北省技术创新专项软科学研究类重点项目"'十五五'湖北基础研究重点方向任务及基础条件自主保障能力研究"(编号:2025EDA011)、青海省重点研发与转化计划"盐湖资源绿色高值利用科技成果数字化服务系统"(编号:2023-QY-213)资助. （编号:2025EDA011）
