基于关系查询的应用系统向语义网移植的研究

针对基于关系数据库的遗留系统能否集成到语义层这一问题,研究关系查询到语义查询转换的完备性.分析一种从关系数据库获取本体的自动学习方法,抽取数据库表、属性同本体类、属性之间映射函数.基于这些函数,得出关系对应的基本简单协议和RDF查询语言(SPARQL)代数表达式.逐一用SPARQL代数来模拟5种基本关系代数运算:选择、投影、并、差和笛卡尔积,并给出其相应的转换算法.结果表明:在这种本体学习的方法下SPARQL代数可以表达5种基本关系代数,以及由它们组合而成的关系表达式,因此,SPARQL是关系完备的,并支持那些基于关系查询的遗留系统向语义层的移植.

Migration of relation query based system to semantic web

Aimed at the problem of whether existing relational database management system(RDBMS) based applications can be integrated to semantic layer or not, the completeness of relation query to semantic query transforming was researched. Based on an Ontology auto learning approach from relational databases, several functions were abstracted to reflect the mappings between relation tables and Ontology classes, as well as relation attributes and Ontology properties. The basic SPARQL algebra expression for relation was established with those mapping functions. Then SPARQL algebras were used to simulate the five basic relational algebra operations: selection, projection, union, difference and Cartesian product, and five corresponding algorithms were proposed. The result showed that the SPARQL algebras can express the five basic relational algebras or the more complex relational algebra expressions composed by them under that ontology learning approach, so the SPARQL is relational completeness, and supports the system migration to semantic layer.