Connected Intelligence

Master 2 – Conformance of RDF data with respect to Web ontologies



In the context of a research contract with ENGIE R&D, we are interested in helping organisations and companies make their systems interoperable on the Web. The RDF data model, and web ontologies, are key enablers for the interoperability on the Web. However, due to the flexibility of the RDF data model, one can use an ontology along with terms from other ontologies, one can encode the same information using different ontologies, one can provide partial information, and all these variations would be considered valid according to the standards. Yet, for best interoperability, information should be provided in forms that are demonstrably processable in a uniform way. Ontologies partly support this but a lot of the requirements that data should fulfil when using an ontology are kept implicit, e.g., in the documentation of the ontology. We need to support developers with tools that can verify if some RDF data conform to ontologies, in order to validate RDF messages with respect to an ontology. There exists no shared understanding of this notion of “conformance”, and some ontology specification documents explicitly include a section about it (e.g., [1]).

A possible, yet rigid, solution is to define the syntactic structures that are allowed to guarantee conformance (e.g., RDF graph patterns). This is the purpose of the SHACL specification that is going to be standardised by the W3C [2]. A more flexible approach should allow using any RDF graph that is logically equivalent to a conforming graph.

We are focusing on using the SEAS modularised and versioned ontologies [3], which are a result of the ITEA 2 European project SEAS (Smart Energy Aware Systems) [4]. These ontologies define various design patterns that may be implemented in new modules, or in external ontologies.

The goal of this internship is first to study how this notion of “conformance” is defined for different existing ontologies, and how it could be defined in a more general way. Then to propose conformance specifications for the SEAS ontologies, and finally to implement these specifications in preliminary tools that will be used by organisations and companies.


Expected results