基于有序聚类的专利知识演化研究

An evolution analysis of patent knowledge based on sequence clustering

据世界知识产权局报告,专利是世界上最大的技术信息源,具有及时、可靠、内容详尽等显著特点,是科技优势的集中体现。目前国内的专利研究主要集中在专利主体上,而专利的文本客体中隐藏了大量的技术信息。利用中国知识产权局专利数据库中汽车专利的标题和摘要两个客体,从中抽取出技术特征,构建专利特征向量,使用有序聚类方法划分国内汽车发展的基本阶段;再利用热点词频和词共现方法分别对划分后的阶段进行分析和对比,揭示每个阶段的技术重心和阶段之间技术重心的变化规律,从技术信息角度构建国内汽车发展的演化过程,为相关从业人员提供参考。

According to the report by the world intellectual property office,patent information is one of the largest scientific and technical information resources with timely,reliable,detailed and other significant characteristics,which can reflection the concentrated advantages in science and technology.However,at present domestic patent research focuses on patent subjects.To find out the hidden technical information in patent texts,we firstly extract the technical features from the title and abstract of car patent from SIPO database,and construct a patent feature vector.Then the orderly clustering method is used to divide the basic stages of domestic car development.Meanwhile,word frequency and word cooccurrence are the two methods used to analyze the differentiated phases,which can provide a better understanding of the gravity of technology as well as the laws for the development of each phase.Finally,we analyze the evolution process of domestic car development from the perspective of technical information,and provide a reference for relevant personnel.