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.
New Approach for the Dynamic Evolution Analysis of Technology Topics:DPL-BMM
Understanding the pulse of technological evolution is important to understand the rules of technological devel-opment.Theme mining based on patent information is an effective way to present macroscopic laws based on the micro-scopic mechanisms of technological development,which are of great significance to technology overlays and innovation-driven 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 technolog-ical 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.
万方数据
维普
