技术主题动态演化分析的一种新方法:DPL-BMM模型  

New Approach for the Dynamic Evolution Analysis of Technology Topics:DPL-BMM

作　　者：宋凯[1] 陈悦[1] Song Kai;Chen Yue(Institution of Science of Science and S&T Management&WISE Lab,Dalian University of Technology,Dalian 116024)

机构地区：[1]大连理工大学科学学与科技管理研究所暨WISE实验室,大连116024

出　　处：《情报学报》2024年第1期25-33,共9页Journal of the China Society for Scientific and Technical Information

基　　金：国家重点研发计划项目“颠覆性技术识别理论、方法与专家预判系统”(2019YFA0707201)。

摘　　要：揭示技术演化脉络是把握技术发展规律的前提,基于专利信息的主题挖掘是基于技术发展微观机制呈现宏观规律的重要研究内容,对技术超前布局和创新驱动实践具有重大意义。技术主题动态演化分析DPL-BMM(Dirichlet process biterm-based mixture model with labelling)是一种附有标签的基于双项狄利克雷过程的混合模型,其突破了传统主题模型在进行主题识别时需固定主题数目的局限,通过增加技术主题表示模块使识别到的技术主题内容更加明确。本文以人工智能领域技术为例进行实证分析,研究结果表明,该方法对技术主题及其演化脉络展示具有实际应用价值。Understanding the pulse of technological evolution is important to understand the rules of technological development.Theme mining based on patent information is an effective way to present macroscopic laws based on the microscopic mechanisms of technological development,which are of great significance to technology overlays and innovationdriven practices.In this study,we propose a model for tracking the dynamic evolution of technology topics based on the DPL-BMM.This model is a Dirichlet process biterm-based mixture model with automatic labeling.This addresses the problem of a fixed number of topics in informatics.A topic representation module was added to identify specific technological topics.The approach was applied to the analysis of patent data in artificial intelligence,and the empirical results show that the method has practical application value for understanding technical topics and their evolution.

关 键 词：技术演化 主题识别 人工智能 可视化 

分 类 号：G301[文化科学] G353.1