Bringing flexibility to ontology learning from relational databases

Current approaches for learning ontology from relational databases are focused on the implementation of the transformation rules without paying attention to user requirements in terms of the way in which the ontology should be designed. They propose ready-made functions that transform the database model to an ontology TBox. There is no way to customize the learning process. The paper proposed a solution based on the transformation of the database model into an XML document, to allow its extension with some meta-data that guide the learning program on how to perform the transformation to ontology. Some case studies demonstrate the effectiveness and efficiency of the proposed approach in terms of reducing the risk that the produced ontology does not fit the requirements of the user. © 2015 IEEE.