基于LDA模型和分类号的专利技术演化研究

发布时间：2018-05-26 18:04

  本文选题：专利文献 + LDA　； 参考：《现代情报》2017年05期

【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。
[Abstract]:Objective / significance: to study the topic evolution of patent literature and analyze the development process and trend of patent technology by using probabilistic subject model. [method / process] LDA model models patent texts according to time window, determines the optimal number of topics by confusion, extracts topic vectors according to the structural characteristics of patent texts, and measures the correlation of topics by JS divergence. This paper introduces IPC classification number to measure the technical topic strength, and finally realizes the evolution research of theme intensity, theme content and technical theme intensity. [results / conclusion] the experimental results show that this method can dig out the subject of patent documents, analyze the evolution law of patent technology over time, and help relevant practitioners to understand the evolution process and trend of patent technology.
【作者单位】： 江西理工大学信息工程学院;
【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)
【分类号】：G255.53
，

本文编号：1938319