基于三维主题特征测度的新兴主题识别研究  

Emerging Topic Recognition Based on Three-Dimensional Topic Feature Measurement

作　　者：郑德俊[1] 程为 Zheng Dejun;Cheng Wei(College of Information Management,Nanjing Agricultural University,Nanjing 210095)

机构地区：[1]南京农业大学信息管理学院,南京210095

出　　处：《情报学报》2024年第2期167-180,共14页Journal of the China Society for Scientific and Technical Information

摘　　要：识别领域新兴主题有利于及时跟踪领域发展的最新动态,为科研工作者的选题以及科研管理者的决策提供情报支撑。本文提出一种基于三维主题特征测度的新兴主题识别方法,基于BERTopic对领域语义知识进行主题建模,以文献为基本单位进行主题表示,构建基于时间、引用和关联的三维主题特征指标框架,用于新兴主题识别;并以文本分类领域为例,验证本文方法的可行性与有效性。研究发现,以文献为基本单位表示主题能辅助主题深入挖掘,三维主题特征指标框架具有较好的适应性与扩展性,本文提出的新兴主题识别方法存在泛化应用的参考价值。在理论层面,能为新兴主题识别的相关研究提供一种可参考的方法和思路;在实践层面,可作为一种参考工具应用于科技情报分析、领域发展态势分析等场景。Recognizing emerging topics is conducive to monitoring the latest trends in development over time,providing valuable information support for researchers’topic selection and research managers’policy decisions.In this study,an emerging topic recognition method based on a three-dimensional topic feature measurement is proposed.First,topic modeling is conducted using domain semantic knowledge from BERTopic,representing topics by documents as the basic unit.Next,a three-dimensional topic feature index framework based on time,reference,and correlation is constructed to identify emerging topics.The feasibility and effectiveness of the proposed method are discussed through empirical studies,using domain data on text classification as examples.The findings reveal that using documents as the basic unit enhances the exploration of topic features,the three-dimensional topic feature index framework demonstrates good adaptability and expansibility,and the proposed method can be generalized application in other domains.At the theoretical level,this work provides a reference method for emerging topic recognition research.At the practical level,it can serve as a reference tool for scientific and technological intelligence analysis and domain development trend analysis.

关 键 词：新兴主题识别 主题建模 主题特征测度 文本分类 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]