Application-oriented purely semantic precision and recall for ontology mapping evaluation

被引：19

作者：

Zhang, Tian ^{[1
]}

Xu, Dezhi ^{[1
]}

Chen, Jianer ^{[1
,2
]}

机构：

[1] Cent S Univ, Sch Informat Sci & Engn, Changsha 410083, Hunan, Peoples R China

[2] Texas A&M Univ, Dept Comp Sci, College Stn, TX 77843 USA

来源：

KNOWLEDGE-BASED SYSTEMS | 2008年 / 21卷 / 08期

关键词：

Application-oriented; Purely semantic; Precision; Recall; Ontology;

D O I：

10.1016/j.knosys.2008.03.060

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 [模式识别与智能系统]; 0812 [计算机科学与技术]; 0835 [软件工程]; 1405 [智能科学与技术];

摘要：

To evaluate and compare various ontology mapping algorithms, it is indispensable to apply them to a group of test ontology pairs and compare the results (named mappings, also known as alignments) with some reference. Currently, the most widely used criteria are precision and recall. However, they do not always work well while facing objects with semantics, like ontologies and alignments, because they only consider "exact" Correspondences but ignore "close" ones. Some new measures have been proposed to solve the problem. However, they have their own limitations, and do not rely purely on the semantics of ontologies and alignments. A framework for purely semantic precision and recall of ontology mapping is proposed in this paper, which can guarantee that the measures are purely semantic. Then a group of application-oriented measures are proposed to instantiate the framework. In some examples these measures are compared with former ones. (C) 2008 Elsevier B.V. All rights reserved

引用

页码：794 / 799

页数：6