Transformation Rules for Building Owl Ontologies from Relational Databases

Abstract

Relational Databases (RDB)

Exploring logical and hierarchical information to map relational databases into ontologies

“…DB2OWL [16] maps tables to classes and columns to predicates, using relational database schema characteristics' to assert sub-classes and object properties. Louhdi et al [6] propose a set of transformation rules which analyzes stored data to detect disjointness and totalness constraints in hierarchies and it calculates the participation level of tables in n-ary relations. Ramathilagam & Valarmathi [4] propose an RDB to Ontol-ogy mapping framework that can generate an Ontology based on the proposed mapping rules for a banking domain.…”

“…Some of the analyzed solutions handles inheritance hierarchies from SQL schemas. They create a subclass hierarchy based on FKs and PKs [12,14,6,15,3,16,17] specifications. However, none of them support the creation of a class hierarchy based on record data extracted from unique columns that have self-hierarchical structures.…”

“…To show how our approach differs from some of the previous works [15,6,4], we compare the resulting ontologies performing a manual mapping using a subset of mapping rules that affects the ontology hierarchical structure (classes and properties). We select these works because they clearly presents the definition and application of their mapping rules, unambiguously.…”

“…Relational databases can be used as the source of information, leading to the creation of systems for instantiating ontologies. The process of building ontologies should be preferably (semi)automated for a knowledge acquisition process that does not have the support of domain experts [6,5]. The RDB structure is intended to be used in different ways in order to construct ontologies, such as simply building an ontology, integrating multiple databases, ontology learning, and managing the organization knowledge [7].…”

“…The works from [9,10,11] provided a classification covering aspects such as the existence of an initial ontology, its application in reverse engineering, automation level, and others. Notably, the approaches [12,13,3,14,15,16,6,17,18,19,5,20] aim to convert relational data models to ontologies, by providing mapping rules that translate the elements of a database schema into concepts and relationships of an ontology. The W3C RDB2RDF Working Group provides as well a survey [21] of different approaches of mapping relational databases to ontologies, including the definition of a reference architecture to compare them.…”