摘要
[目的/意义]通过对美国空军公布的专利进行研究,揭示其不同阶段的重点研究方向,更好地把握技术变化,为相关决策提供参考。[方法/过程]引入时间维度将主题模型生成的"文档—主题"分布转换为"年份—主题"分布,结合t-SNE降维并可视化来划分时间段。采用专利分类号与技术领域的映射关系和社会网络分析方法做粗粒度的技术领域演化及可视化分析,并综合运用LDA和JS散度对其中某一技术领域做细粒度的技术主题演化与可视化分析。[结果/结论]将美国空军公布的1958—2019年的专利划分为7个时间段,得到技术领域演化河流图以及"测量"技术领域的技术主题演化桑基图。结果表明该方法可以较好地识别美国空军不同阶段的核心技术演化过程,是从专利的角度研究国防技术发展趋势的一次积极探索。
[Purpose/significance]By studying the patents published by the United States Air Force,it reveals the key research directions at different stages of national defense to better grasp the technological changes and provide a reference for relevant decisions.[Method/process]The"document-topic"distribution generated by the topic model is transformed into a"year-topic"distribution by introducing a time dimension,which is reduced and visualized by the t-SNE algorithm to divide the time period.The mapping between IPCs and technology fields and the method of complex network analysis are used to do the coarse-grained technology field evolution and visualization analysis,and the fine-grained technology theme evolution and visualization analysis of one of the technology fields is done by combining LDA and Jensen-Shannon divergence.[Result/conclusion]The U.S.Air Force’s published patents from 1958-2019 are divided into 7 time periods to obtain an evolutionary themeriver diagram of the technology fields and a Sankey diagram of the evolution of technology themes in the"Measurement"technology field.The results show that this method can better identify the core technology evolution process in different stages of the U.S.Air Force,which is an active exploration of the development trend of national defense technology from a patent perspective.
出处
《情报理论与实践》
CSSCI
北大核心
2021年第2期41-49,共9页
Information Studies:Theory & Application
基金
中央高校基本科研业务费专项资金项目“专利视角下的技术创新主题发现与趋势预测”的研究成果之一,项目编号:JB190610。
