基于主成分分析和K-Modes蚁群聚类的本体映射方法  

ONTOLOGY MAPPING METHOD BASED ON PRINCIPAL COMPONENT ANALYSIS AND K-MODES ANT COLONY CLUSTERING

作　　者：叶霞 许飞翔 曹军博 王馨 Ye Xia;Xu Feixiang;Cao Junbo;Wang Xin(Academy of Combat Support,Rocket Force University of Engineering,Xi an 710025,Shaanxi,China;Department of Information Systems,University of Maryland,Baltimore,Baltimore 21250,Maryland,USA)

机构地区：[1]火箭军工程大学作战保障学院,陕西西安710025 [2]马里兰大学巴尔的摩校区信息系统学院,马里兰州巴尔的摩21250

出　　处：《计算机应用与软件》2020年第12期231-237,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(61702525)。

摘　　要：指挥信息系统测试数据在利用本体技术进行数据集成时,现有的本体映射模型准确率不高,影响集成效果。针对这种情况,提出基于主成分分析(Principal Component Analysis,PCA)和K-Modes蚁群聚类(K-Modes Ant Colony Clustering,KMACC)算法的本体映射方法。在基于PCA算法的本体概念相似度综合计算模型基础上,引入KMACC算法,实现批量本体概念映射关系的发现,进一步提高映射的准确性。实验结果表明,相比于传统基于Hownet的映射方法和RiMOM方法,该方法有效提升了本体映射的查全率和查准率,较好地解决了本体概念集成中的关键问题。When the test data of the command information system is integrated by using ontology technology,the accuracy of existing ontology mapping model is not high,which affects the integration effect.In view of this,we propose an ontology mapping method based on principal component analysis(PCA)and K-modes ant colony clustering(KMACC)algorithm.On the basis of PCA-based comprehensive calculation model of ontology concept similarity,the KMACC algorithm was introduced to realize the discovery of ontology concept mapping relationship in batch,and further improve the accuracy of mapping.The experimental results show that compared with the traditional Hownet-based mapping method and RiMOM method,our method effectively improves the recall and precision of ontology mapping,and better solves the key issue in ontology concept integration.

关 键 词：本体映射 主成分分析 相似度计算 K-Modes聚类 蚁群聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]