结合属性分布特征的模式匹配算法

该文提出了一种结合属性分布特征的Web模式匹配算法,属性分布特征包括属性对互斥特征和属性对共现特征。属性对互斥特征由属性对的互斥性和出现次数计算得出,这个特征隐含了属性对的语义相似程度。为了充分利用传统的属性名、属性值相似性特征,该文通过机器学习方法结合属性对互斥特征与相似性特征进行属性匹配。并以潜在的匹配属性对为基础,引入有约束的属性聚类方法进行Web模式匹配,聚类方法的约束条件来自属性对共现特征。实验结果表明,相对于仅使用相似性特征的方法,在不同的实验设置下,结合属性分布特征的Web模式匹配算法将F值提高了0.13到0.55。

Ontology-based schema matching method in web query interface integration

Deep web data integration needs to do schema matching on web query interfaces and obtain the mapping table.By introducing semantic conflicts into web query interface integration and discussing the origins and categories of the semantic conflicts,an ontology-based schema matching method is proposed.The process of the method is explained in detail using the example of web query interface integration in house domain.Conflicts can be detected automatically by checking semantic relevance degree,then the categories of the conflicts are identified and messages are sent to the conflict solver,which eliminates the conflicts and obtains the mapping table using conflict solving rules.The proposed method is simple,easy to implement and can be flexibly reused by extending the ontology to different domains.