基于GRA本体映射研究

Ontology Mapping Based on GRA

基层用户产生的本体映射关系蕴含着相关领域内的类层次结构,利用这部分知识可用于新映射关系的产生.但是现有系统存在以下缺点:类层次复杂性考虑不足,矛盾知识不完全删除和效率低下.针对这些问题,文章提出将基层本体映射的相似和等价关系进行不同处理的方法,并重新设计了类层次知识库的前向推理规则.实验表明,改进的系统使效率得到提高,在保证查全率的同时可以明显改善查准率,而且能较好支持本体复合映射.

Ontology mapping created by endusers contains class hierarchy structure knowledge of the domain, and using that knowledge could produce new domainal ontology mapping. However, current systems have the following drawbacks： insufficient consideration of class hierarchy complexity, incompletely deleting inconsistent knowledge and inefficiency. Aiming at solving those problems, this paper proposes a new method of dealing differently with ＂similar to＂ and ＂equal to＂ relations in grass-roots ontology aligm^ent. Besides, this paper redesigned inferences on class hierarchy knowledge base. The experiment result shows that revised system is more efficient than the original one, can produce results with higher precision while maintaining high recall, and can support complex ontol- ogy mapping.