Uma abordagem para publicação de visões RDF de dados relacionais

Abstract

The Linked Data initiative brought new opportunities for building the next generation of Web applications. However, the full potential of linked data depends on how easy it is to transform data stored in conventional, relational databases into RDF triples. Recently, the W3C RDB2RDF Working Group proposed a standard mapping language, called R2RML, to specify customized mappings between relational schemas and target RDF vocabularies. However, the generation of customized R2RML mappings is not an easy task. Thus, it is mandatory to define: (a) a solution that maps concepts from a relational schema to terms from a RDF schema; (b) a process to support the publication of relational data into RDF, and (c) a tool that implements this process. Correspondence assertions are proposed to formalize the mappings between relational schemas and RDF schemas. Views are created to publish data from a database to a new structure or schema. The definition of RDF views over relational data allows providing this data in terms of an OWL ontology structure without having to change the database schema. In this work, we propose a three-tier architecture – database, SQL views and RDF views – where the SQL views layer maps the database concepts into RDF terms. The creation of this intermediate layer facilitates the generation of R2RML mappings and prevents that changes in the data layer result in changes on R2RML mappings. Additionally, we define a three-step process to generate the RDF views of relational data. First, the user defines the schema of the relational database and the target OWL ontology. Then, he defines correspondence assertions that formally specify the relational database in terms of the target ontology. Using these assertions, an exported ontology is generated automatically. The second step produces the SQL views that perform the mapping defined by the assertions and a R2RML mapping between these views and the exported ontology. This dissertation describes a formalization of the correspondence assertions, the three-tier architecture, the publishing process steps, the algorithms needed, a tool that supports the entire process and a case study to validate the results obtained.