In this work, we describe an approach for mapping the structure of a database schema to the structure of existing ontologies. The database contains plant traits values and plant experimental history. The goal is to identify the semantic correspondences between databases and ontologies and provide a tool that can be more broadly adopted by the community. The approach presented in this work is a semiautomatic approach. In the literature, there are several approaches which map the database schema to an ontology. The underlying assumption by all approaches is that the chosen ontologies model the same domain as the one modelled by the relational database schema. Some mapping approaches are R2O (Barrasa et al., 2004), DartGrid (Chen et al., 2006), Linked Data Mapper (Zhou et al., 2008), RDOTE (Vavliakis et al., 2010), RDB2OWL [Bumans & Cerans, 2010], MAPONTO [An et al., 2006]. The difference between the above approaches is that some approaches are manual and some are semimanual. Furthermore, for some approaches a human expert gives the correspondences between database terms and ontology terms.
The database schema of our Phenomis database considered for mapping contains plant phenotyping information and environmental information. The ontologies considered for mapping are plant ontology, phenotypic quality ontology, plant trait ontology, plant environmental conditions, and environment ontology. The mentioned ontologies are very large and contain over 1000 concepts. Unlike the number of concepts, the number of roles is very small (less than 10). Note that the roles denote the relations between domain objects.
In order to map the database to the ontology, we first
consider the schema of the database and extract relation names
and attributes of each relation. Note that the relations in a
relational schema are classified into two categories, namely
entity relations and relationship relations