基于主题模型和专利数据的技术创新评价研究  

Research on Evaluation of Technical Innovation Based on Topic Model and Patent Data

作　　者：余本功[1,2] 陈杨楠 杨颖[1,2] Yu Bengong;Chen Yangnan;Yang Ying(School of Management, Hefei University of Technology, Hefei 230009, China;Key Laboratory of Process Optimization & Intelligent Decision-making, Ministry of Education,Hefei University of Technology, Hefei 230009, China)

机构地区：[1]合肥工业大学管理学院,安徽合肥230009 [2]合肥工业大学过程优化与智能决策教育部重点实验室,安徽合肥230009

出　　处：《现代情报》2019年第1期111-117,168,共8页Journal of Modern Information

基　　金：国家自然科学基金项目"基于制造大数据的产品研发知识集成与服务机制研究"(项目编号:71671057);"不确定环境下的复杂产品研发协同绩效动态评价研究"(项目编号:71573071);过程优化与智能决策教育部重点实验室开放课题资助

摘　　要：[目的/意义]专利是企业技术创新活动的重要成果,对专利数据进行分析,有利于客观评价企业技术创新能力。[方法/过程]从计量的角度对企业专利数据进行分析的同时,结合机器学习的方法,通过LDA模型对专利摘要文本进行内容挖掘,构建基于专利文本内容的评价指标,建立由专利数量、专利趋势和专利内容三方面指标组成的技术创新评价体系。[结果/结论]采用熵值法确定各项指标对企业技术创新的影响权重,并通过实验对国内自主品牌制造企业进行技术创新评价,说明了评价方法的现实意义。[Purpose/Significance]Patent is an important achievement of technical innovation activities in enterpri.ses,which being analyzed is conducive to evaluating the technological innovation ability of enterprise objectively.[Method/Process]Analyzed the enterprise patent data from the perspective of measurement,combined the methods of machine learning to carry out content mining of patent summary text through LDA model to construct evaluation indicators based on patent text content,and established a technical innovation evaluation system based on the number of patents,patent trends and patent content.[Result/Conclusion]Entropy method was used to determine the indicators weight influence on domestic manufacturing enterprises.It was proved that practical significance of the evaluation method was to evaluate the capability of the domestic manufacturing enterprise by experiment.

关 键 词：主题模型 专利数据 LDA 内容挖掘 熵值法 技术创新评价 

分 类 号：G306[文化科学]