BP. 5: Define crosswalks between different formats. (D2.5)

[GCoen1]

Semantic artefact can be serialized in various formats (SKOS, RDF, OWL, XML, …). This diversity of formats makes it complicated to integrate and work with heterogeneous semantic artefacts. Practitioners should describe or reference the particular crosswalk they used to convert the semantic artefact from one format to another. These workflows could be defined using machine readable mappings.

[matentzn]

This is interesting, could we get some clarifications here? There are two level I think that should be distinguished. On the serialisation level, we could have an (OWL) ontology being exported as RDF/XML (with owl constructs), Functional syntax or even OBO (if one considers that a "syntax" of OWL, which some do). On the formalism level this is much harder. Say you have

my:Diabetes skos:broader my:Disease

How would you translate that to OWL? You could of course pretend that my:Diabetes and my:Disease are classes and assert an annotation property skos:broader between them, but what about the way back?

my:Diabetes rdfs:subClassOf my:Disease

You could translate it into the skos:broader statement above, but not sure this makes sense to that in all cases.

The question is: Is this ticket about serialisations or mappings between formalisms?

[luizbonino]

Hi @matentzn , this ticket is about serialisation. The issue on formalism mapping is also very interesting and complex. I agree with you that, in principle, subclassing implies that the super class is broader than the sub class. But also depends on the precise semantics of "broader". The regular definition of a sub class is that it specialises the super class, adding further constraints. In other words the sub class is narrower than the super class because for an entity to be classified as the sub class, it has to go through more strict (narrower) criteria than the super class. Unfortunately, the skos:broader does not have very precise semantics. E.g., how much broader a class has to be in relation to other to be considered skos:broader and not a skos:closeMatch?

Coming back to the original question, this issue is about serialisation. One of the properties for a semantic artefact is the serialisation format(s) used to represent it. And since some of the available formats can be converted from one format to another, the definition of these crosswalks would be useful.

[matentzn]

@luizbonino thank you about the clarification!