基于模式结构分类的本体映射方法

摘要

Ontology mapping plays a key role in semantic integration area,a top-down ontology mapping method is proposed by this paper, it considering four basic mapping and this mapping process is assumed that all cases are 1: 1. This approach describes the ontology as a graph and each ontology concept defined as a tree,on the basis of the tree,give the similarity computation method of leaf nodes and non-leaf nodes,merge the concept sub-tree by concept classification,According to the results of classification reorganization ontology chart structure. Finally, give a complete ontology mapping model. If the number of ontology is greater than two,concept of group separation method will be used,that is,a similar ontology in the same group,no similar in a different group,the separation date until convergence. Experimental results show that the layer classification ontology mapping method can reducenon-related concepts among count for large scale structure ontology mapping for the efficiency and accuracy it achieve the desired results.%本体映射是语义集成的关键技术,本文基于本体的模式结构提出了一种自顶向下的本体映射方法,该方法在考虑四种基本映射关系并假定所有映射关系均为1:1的情况下,首先将本体描述为图,并将本体中的每个概念都定义为树,在此基础上给出了树中叶节点、非叶节点相似度的计算方法,通过概念分类将子树进一步合并,根据分类的结果重新组织图结构,最后给出了一个完整的本体映射模型.当本体的数目大于2个时,采用概念组分离的方法,即相似的本体在同一组中,不相似的本体在不同的组中,分离直至收敛为止.实验结果表明,这种自顶向下逐层分类的本体映射方法在对大规模结构化本体进行映射发现时可以有效地减少不相关概念之间的计算,在效率和准确度上均取得了理想的效果.