计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (20): 33-36.DOI: 10.3778/j.issn.1002-8331.2010.20.009

• 研究、探讨 • 上一篇    下一篇

多策略概念相似度计算方法LMSW

夏红科1,2,郑雪峰1,胡 祥3   

  1. 1.北京科技大学 信息工程学院,北京 100083
    2.北京信息科技大学 计算机学院,北京 100192
    3.华北电力大学 计算机科学与技术学院,北京 102206
  • 收稿日期:2009-01-20 修回日期:2009-03-23 出版日期:2010-07-11 发布日期:2010-07-11
  • 通讯作者: 夏红科

Multi-strategy concept similarity computing method LMSW

XIA Hong-ke1,2,ZHENG Xue-feng1,HU Xiang3   

  1. 1.Information Engineering Institute,University of Science and Technology Beijing,Beijing 100083,China
    2.Computer School,Beijing Information Science and Technology University,Beijing 100192,China
    3.Computer Science and Technology School,North China Electric Power University,Beijing 102206,China
  • Received:2009-01-20 Revised:2009-03-23 Online:2010-07-11 Published:2010-07-11
  • Contact: XIA Hong-ke

摘要: 本体映射是语义集成的关键,国内外对本体映射进行了深入的研究。将机器学习的方法引入到组合概念相似度的权值计算中来,提出了基于Widrow-Hoff理论的权值确定算法LMSW。通过梯度下降法来搜索最佳拟合训练样例的权向量,从而计算出组合的概念相似度,最终实现组合多种概念相似度算法的本体映射。实验表明该算法比单独使用概念相似度计算方法在查全率、查准率上均有所提高。

关键词: 本体映射, 概念相似度, 权值分配

Abstract: As a crucial point of semantic integration,ontology mapping is researched in-depth at home and abroad.The machine learning method is introduced to weights computing in combining concept similarities,and the weight allocation algorithm LMSW based on Widrow-Hoff theory is proposed.In LMSW,the weight vector most fitting the training samples is searched by gradient descent method,so that the final combined concept similarity can be attained,and the ontology mapping which combines multiple concept similarity strategies can be determined eventually.Experiment indicates that the method performs well in improving precision and recall compared to individually using other algorithms.

Key words: ontology mapping, concept similarity, weight allocation

中图分类号: