START Conference Manager    

Transforming UML Application Schema to Linked Data Ontologies. (A Dutch best practice)

Paul Janssen, Linda van den Brink and Marco Brattinga

(Submission #113)


Abstract

The use of linked data as an alternative encoding for UML described domain models is becoming more and more a common practice. In the Netherlands the national standard for geo-information modelling prescribes UML as the standard for specifying conceptual domain models and GML as the preferred encoding using the ISO 19118 and ISO 19136 as the UML – GML encoding rules. It also states that when a different encoding is used proper encoding rules must be referenced or be developed. For UML to RDF several specifications of encoding rules exist including major contributions by INSPIRE , INSPIRE Guidelines for RDF encoding for spatial data. However, these encoding guidelines indeed are guidelines and can only be implemented after a detailed fit for purpose analysis. This leading to more detailed rules described by best practices.

A team was established representing the Dutch geo information modelling domain both in UML object oriented data as well as three main Geo-RDF implementation approaches. The latter being COINS for the exchange of BIM information; OrOx as a RDF vocabulary for Sewer Utility Network information and a generic national implementation of W3C/SHACL RDF. The approach centred around developing transformation rules for mapping from the source UML geo metamodel to the target metamodel(s) for the SKOS/RDF/SHACL/OWL implementations. The rules were developed on the basis of the existing metamodels but also on a use case about a test case domain model: the Golf Course Domain Model and a related dataset.

The research showed that transformation rules can be developed but that an automated-only transformation is not a preferred solution and will lead to sub-optimized ontologies. Automated transformation can lead to a valid model that follows all the rules, but not as a correct representation of the Universe of Discourse. It is therefore recommend to manually design and optimize RDF ontologies from automated transformation of source UML domain models guided by a combination of both predefined transformation rules as well as the understanding of the Universe of Discourse. We have identified a set of best practices and common pitfalls that occur when blindly transforming a UML model tot RDF. This intelligence guided approach is described by a set of the best practice transformation rules. The hands-on project strengthened the collaboration between two still different geo-implementation communities and shows how data and models can be connected

Categories

Topic Area:  [3.5] Best practices
Abstract Type:  Oral Presentation

Additional Fields

 
Academic:   Yes
 
Data Provider:   No
 
Data User:   No
 
INSPIRE Implementer (IT):   Yes
 
INSPIRE newbies:   No
 
Policy Officers:   No
 
Public Administration (MS/Regional/Local):   No
 
Thematic specialists:   No
 
Other_theme:   Linked data community, datamodeller
 
Comments:   UML model, RDF Ontologies, OWL, Linked data


START Conference Manager (V2.61.0 - Rev. 5269)