基于重构向量空间模型的知识匹配算法研究  

Knowledge Matching Algorithm Based on Reconstructed Vector Space Model

作　　者：黄振峰[1] 刘皓天 吴振勇[1] 武羽馨 HUANG Zhen-feng;LIU Hao-tian;WU Zhen-yong;WU Yu-xin(College of Mechanical Engineering,Guangxi University,Guangxi Nanning530004,China)

机构地区：[1]广西大学机械工程学院

出　　处：《机械设计与制造》2020年第2期203-206,共4页Machinery Design & Manufacture

基　　金：广西自然科学基金（No.2016GXNSFBA380184)

摘　　要：针对产品设计知识管理系统中基于向量空间模型的知识匹配算法准确率和召回率不高的问题,通过研究产品设计知识的特点、传统向量空间模型存在的缺陷,重构向量空间,建立语义及语义结构向量空间模型,在此基础上提出了一种基于重构向量空间模型的知识匹配算法。针对该算法,分别研究了特征项权重计算、特征项相似度算法,引入了文本语义结构相似度,通过文本相似度计算对设计人员匹配适当的知识,开发了知识管理原型系统,并以某环保企业的产品设计知识为例,验证了该方法具有较好的准确率、召回率以及F1-score。Aiming at the problem that the accuracy and recall rate of knowledge matching algorithm based on vector space model in product design knowledge management system are not high,the characteristics of product design knowledge and the defects of traditional vector space model were studied,and the vector space was reconstructed,then the semantics and semantics structure vector space model was established,on the basis of which a knowledge matching algorithm based on reconstructed vector space model was proposed.Aiming at the algorithm,the feature weight calculation and feature similarity algorithm were studied respectively,and the similarity of text semantic structure was introduced,and the appropriate knowledge is matched for the designer by text similarity calculation.The knowledge management prototype system was developed and the product design knowledge of an environmental protection enterprise was taken as an example to verify that the method has good accuracy rate,recall rate and F1-score.

关 键 词：知识管理 知识推送 匹配算法 文本相似度 向量空间模型 

分 类 号：TH16[机械工程—机械制造及自动化] TH12