A comparison of information modeling in ORM and OWL

by Charlotte Løvdahl

The Semicolon II project looks at ways to enhance the interoperability between Norwegian government institutions by using metadata and conceptual models. The Web Ontology Language (OWL) has been chosen to create the metadata models for these institutions. This thesis focuses on comparing Web Ontology Language (OWL) and Object Role Modeling (ORM). ORM is a well established modeling language in comparison to OWL which is relatively young. Their capability and suitability to create information models are investigated with a mapping from ORM to OWL. This mapping revealed that OWL can construct most of the constraints in ORM, except for the Equivalence-Of-Path constraint. The Open World Assumption in OWL complicates information modeling by not forcing restrictions on classes. This allows for incomplete information: information which is expected, but not found, is stated as unknown. Thus a model can violate the rules on which it is built. How to deal with this disadvantage is a topic for future research. The languages ability to support and underpin the information systems they model are investigated by researching tools that combine data from databases with the models. There are, to our knowledge, no tools for ORM which allows its models to connect to data in a database, while OWL has several. Based on a survey of available tools, we have chosen to apply the tool D2RQ. The result is an OWL model with a bridge to the data in the database. This allows the information system to use the database's strengths on query performance and storage. In addition the populated model can be reasoned over and put on the Web for easy sharing.

The entire thesis is available at this site.

On the right hand side you see the links for the documents created for this thesis. The second to last link is the server that runs the mapping from the database. The last link is the OWL model populated with the data from the database.