基于降维理论的本体映射算法  

Ontology mapping algorithm based on dimensionality reduction theory

作　　者：徐天伟[1] 甘健侯[1] 高丽金[1] 高炜[1,2] 

机构地区：[1]云南师范大学信息学院,云南昆明650092 [2]苏州大学数学科学学院,江苏苏州215006

出　　处：《西北师范大学学报（自然科学版）》2011年第4期52-55,共4页Journal of Northwest Normal University(Natural Science)

基　　金：国家自然科学基金资助项目(60903131)

摘　　要：通过建立多本体间的映射,实现多本体间的相互操作.将多本体图中每个顶点对应的信息用一个向量表示,通过计算多本体图中每对顶点的欧式距离建立一个连通的本体图,利用谱聚类降维方法将本体图中每个顶点的数据信息降为一维,从而通过顶点对应的实数间的差值判断来自不同本体顶点间的相似程度,进而建立本体映射.实验结果表明,新算法能较好解决已有文献中存在的问题,在结构存在差异的本体间建立映射是有效的.This paper aims to establish a mapping by multiple ontologies to achieve interoperability between multiple ontologies.Each vertex in multiple ontology graphs can use a vector corresponding to the vertex＇s information.The connected ontology graph is obtained by calculating the Euclidean distance for each pair of vertices.Using spectral clustering dimension reduction methods,the dimension for each vertex is reduced to one-dimensional data.The ontology mapping is given by measuring the difference of values which reflect their similarities.Experimental results show that the new algorithm can solve the problem of the literature and is effective for mapping in ontologies of large differences.

关 键 词：本体映射 概念相似度 降维理论 欧式距离 拉普拉斯Beltrami算子 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]