摘要
[目的 /意义]运用概率主题模型全面研究专利文献主题演化,分析专利技术发展过程及趋势。[方法/过程]LDA模型按时间窗口对专利文本建模,困惑度确定最优主题数,按专利文本结构特性提取主题向量,采用JS散度度量主题之间的关联,引入IPC分类号度量技术主题强度,最后实现主题强度、主题内容和技术主题强度3方面的演化研究。[结果 /结论]实验结果表明:该方法能够深入挖掘专利文献的主题,可以较好地分析专利技术随时间的演化规律,帮助相关从业人员了解专利技术的演化过程及趋势。
[Purpose/significance] This paper used the probability topic model to study the evolution of patent literature, and analyzed the development process and trend of patent technology. [ Methods/process] The LDA model modeled the patent text by time window, confusion degree method was used to determined the optimal number of topics, extracting topic vectors according to the structural characteristics of patent text, used the JS divergence to measure the association between topics, introduced IPC classification number to measure technical topic strength, finally, the evolution of topic strength, topic content and technical topic strength were studied. [Results/conclusion] The experimental results showed that this method could deeply excavate the topic of the patent literature, and could analyze the evolution of patent technology over time and help the practitioners to understand the evolution process and trend of patent technology.
作者
廖列法
勒孚刚
Liao Liefa Le Fugang(School of Information Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China)
出处
《现代情报》
CSSCI
北大核心
2017年第5期13-18,共6页
Journal of Modern Information
基金
国家自然科学基金项目"创新网络异质性与企业创新绩效关系研究"(项目编号:71462018)
江西省研究生创新专项基金资助项目"基于领域知识的LDA主题模型"(项目编号:YC2015-S304)
关键词
专利文献
LDA
JS散度
IPC分类号
技术主题强度
专利技术演化
patent literature
LDA
JS divergence
IPC classification number
technical topic strength
patent technol- ogy evolution
作者简介
廖列法（1975-），男，教授，博士，研究方向：电子商务、大数据分析。
