基于多维特征相似度的专利预警分析研究  

Patent Warning Analysis Based on Multidimensional Feature Similarity

作　　者：林群庚 王薄雯 赵乃萱 邓一默 杜晓月 LIN Qungeng;WANG Bowen;ZHAO Naixuan;DENG Yimo;DU Xiaoyue(School of Business Administration,Northeastern University,Shenyang Liaoning 110167,China;School of Software,Northeastern University,Shenyang Liaoning 110167,China)

机构地区：[1]东北大学工商管理学院,辽宁沈阳110167 [2]东北大学软件学院,辽宁沈阳110167

出　　处：《信息与电脑》2023年第13期97-100,共4页Information & Computer

摘　　要：专利是科技创新成果的重要载体,如何有效地对其进行保护、分析与运用始终是各界普遍关注的问题。专利预警分析是基于专利信息对技术领域或行业的竞争态势等进行分析和预测的方法,对专利权人保护知识产权、提高创新能力等方面具有重要意义。现有专利预警分析方法普遍存在效率低、精度差、覆盖面小等问题。为此,提出了一种基于多维特征相似度的专利预警分析方法,通过Doc2Vec、Jaccard系数及熵权法获得整体相似度,进而构建相似专利知识图谱实现可视化建模。Patent is a significant carrier of scientific and technological innovation,how to effectively protect,analyze and apply it has always been a common concern.Patent warning analysis is a method to analyze and predict the competition situation in the technology field or industry based on patent information,which is of great significance to the patentee's protection of intellectual property rights and improvement of innovation ability.Existing patent warning analysis methods generally have problems such as low efficiency,poor accuracy and small coverage.Therefore,this paper proposed a patent early warning analysis method based on multidimensional feature similarity,obtained the overall similarity through deep learning Doc2Vec and Jaccard coefficients,and then constructed a similar patent knowledge map to realize visual modeling.

关 键 词：专利预警分析 Doc2Vec Jaccard系数 熵权法 知识图谱 

分 类 号：G251.2[文化科学—图书馆学]