本体辅助的自动化模式匹配技术

Ontology-Aided Automatic Schema Matching

在基于映射的数据交换系统框架下,提出了一种本体辅助的模式匹配方法.它利用WordNet词汇本体和决策树学习相结合的方法进行属性名称匹配,构建数据类型本体计算属性数据类型的语义距离,依赖领域本体发现一对多的语义匹配关系,这3个过程逐步提高了匹配质量.建立在实际应用数据上的实验结果表明,该方法具有较高的精确度和召回率.

This paper introduces an ontology-aided schema matching method in the mapping-based data exchange framework. The decision tree learning and WordNet are used to match attribute names. A data type ontology is constructed to compute semantic distance between attribute data types. Also, domain ontologies can be used to detect l：n semantic matches. The three steps improve the match quality steadily. Experiments of several real applications show encouraging results, yielding high precision and recall measures.