基于实例的语义相似属性识别方法

为了解决不同数据源集成时的语义冲突,实现语义共享,首先必须要进行语义相关属性识别.针对不同类型的语义相关属性及其实例的表达形式,提出了一种更加具有通用性的语义相关属性识别方法--属性相似性计算模型,该模型能够计算数值型属性、字符型属性、标称型属性等三种不同类型属性之间的相似性,为判断不同属性是否相关提供了量化依据.结果表明,所提出的属性相似性计算方法不仅可以有效计算属性的相似性,而且允许相关属性的实例值存在表达冲突,拼写错误等情况,因此,该方法具有较好的容错性和通用性.

Instance-based approach to identify semantically similar attribute

To resolve semantic conflicts of heterogeneous data sources integration and realize semantic sharing,semantic related attribute should be firstly identified.For different kinds of attributes,a more general instance-based approach to identify semantically related attribute---attribute similarity calculating model is put forward,which could be used to compute similarity value of three different kind of attributes including numerical attributes,character attributes,and coded attributes,and provide quantitative criteria to decide whether two attributes is semantically related.The result shows that the presented model not only compute attribute similarity effectively,but also allow instance of related attributes to be expressed differently or have a little spelling mistake,therefore,this method is more error-accepted and general.