基于类比设计的专利知识挖掘算法

摘要

To solve the problem that the inventive patent was hard to realize breakthrough and innovation caused by lacking scientific theoretical system and enough technical paper,a patent mining algorithm based on analogy design was suggested.The mature technology patents with similar properties or functions were taken as the source data to do the text mining.The Problem Solution Concepts (PSCs) was extracted from source data,and the Vector Space Model (VSM) based on PSCs' TF-IDF values was created.According to the similarities among patents,PSC-based patent map was constructed.The opportunity score was calculated and the PSC which had a high opportunity score was chosen to create a new innovation scheme on the new technology with analogy.The architecture design of patent mining system was given,and its dynamic executive process was described with Unified Modeling Language (UML).The data of system was stored in MySQL,programed by JAVA and tested by a case study of wireless charger technology and wireless sensing technology.The experiment result showed that the proposed algorithm was more effective and could create a more innovational scheme than what the comparative algorithm could do.%为解决发明专利因缺少完善的理论体系和相关技术文献而难以实现技术突破和创新的难题,提出一种基于类比设计的专利挖掘算法.以具有相同特征和相似功能的成熟技术专利为数据源进行文本挖掘,首先,从数据源中提取实现功能和问题解决方案(PSC),建立基于PSC的词频—逆向文件频率(TF-IDF)值的向量空间模型;然后,根据专利文献间的相似度制成基于PSC的专利地图;最后,分别对PSC进行可行性分析,根据分析结果选择合理的方法,用类比的思想进行技术创新.给出了专利知识挖掘应用系统的架构设计,并采用统一建模语言描述了专利挖掘系统的动态执行过程,获得了专利挖掘系统的时序图.应用系统采用关系型数据库MySQL存储数据,运用Java编程语言开发了应用系统,以无线充电技术专利和无线传感技术专利为源数据测试系统.实验结果表明,所提算法比基于专利的类比方法具有更好的时间性能,并能获得更具创新性的创新肯案.