基于向量空间的专利类比挖掘算法

摘要

It is hard to define problems accurately and figure out solutions in time when the innovation needs to make a technological breakthrough.To remedy existing limitations,patent mining algorithm based on VSM using analogy is suggested.Firstly,extract Problem Solution Concepts (PSCs) from source data and create Vector Space Model (VSM) based on TF-IDF values of PSCs.Then,construct patent map according to the similarities among patents.Finally,calculate the opportunity score and choose the PSC which has a high opportunity score to create new innovation scheme on the new technology using analogy.The time complexity of the suggested algorithm is O [(n2 + n)/2] which is lower than O (rt2) that belongs to comparative algorithm.The algorithm is tested by a case study of wireless charger technology and wireless sensing technology.It turns out that the suggested algorithm is more capable of creating innovational scheme than the comparative algorithm.%新技术在进行技术突破时,很难找到创新点和方案.针对这一难题,提出一种基于向量空间的专利类比挖掘算法.首先,从源数据中获取描述功能和属性的专利技术方案(PSC),建立基于PSC的TF-IDF值的向量空间模型(VSM);然后,根据专利文献间的信息距离制成基于PSC术语的专利地图;最后,分别对PSC进行创新性分析,根据分析结果用类比的思想进行新技术的创新.本文算法的时间复杂性为O[(n2+n)/2],低于对比算法的O(n2).以无线充电技术专利和无线传感技术专利文献为源数据,实验结果表明,所提出的算法比对比算法能更有效地获取更具有创新性的创新方案.