I think this is a good idea. However, I would be more specific about what compliance means. Perhaps the most future proof is something like: "The ontology must be valid OWL DL as determined by the Manchester owl Lint plugin to Protege 4." Probably this is equivalent to requiring determination by the Manchester WonderWeb OWL validator.
The biggest recurring problem I've had with identifying the OWL species of TDWG ontologies is that frameworks try to resolve some of the legacy DC terms to the old versions of DC, many of which are not OWL DL for relatively simple reasons. Try, e.g. http://rs.tdwg.org/ontology/voc/SpeciesProfileModel.rdfhttp://rs.tdwg.org/ontology/voc/InstitutionType.rdfin WonderWeb, which quickly indicts some older DC that can easily be replaced with the 2008 versions. (It also finds some similar lint in TDWG owl files).
Aside from that, I remain silent, in agreement with Gregor.
On Tue, May 12, 2009 at 4:59 AM, Roger Hyam rogerhyam@mac.com wrote:
Hi All, I need to do some work on the Taxon Name and Taxon Concept vocabularies and believe I have come up with a good way of organising the TDWG ontology space (everything within http:/rs.tdwg.org/ontology).
The following are the changes I suggest:
- All files should be OWL DL compliant
- All files should be openable in Protege 4 (I believe this is now good
enough to use for editing these small ontologies)
- We take a highly structured modular approach I call this the Bricks
and Mortar design pattern - Some files are 'Bricks' and as such *import or reference no other files, classes or individuals*. e.g. TaxonName does not mention a higher 'Name' object in the class hierarchy. - Other files are 'Mortar'. These files import Bricks and stipulate relationships between things. Because we are using OWL it is easy to define things like the class hierarchy or the range of a property in a separate file to the file the original class or property was defined in. - This pattern gives us maximum re-usability as the same Brick could be used in different ways. It does not bind us to any one implementation of one object. - An example of the usage pattern would be to define TaxonName, TaxonConcept, Rank, NomenclaturalCode as separate bricks that don't reference each other at all then create a TCS ontology that imports these 4 bricks and defines their relationships.
- We move to some other method of presenting the ontologies on line -
possibly the OWLDoc plug-in for Protege. This would lose us the branded look we have at the moment but would be more flexible and consistent in the long run.
As I need to do this for the TaxonName TaxonConcept vocabularies I volunteer to do manage the space this year if people are happy going down this route.
From the point of view of deployed systems (the nomenclators) there may be a need for a namespace change on some properties but I would review what is in use and this would be trivial - if necessary at all.
What do you think? I will take silence as acquiescence on the grounds that any movement is better than none - though I don't suppose I will get round to doing anything about changes till after e-Biosphere in June.
All the best,
Roger
Roger Hyam - Project Officer WP4 Pan European Species Infrastructure
+44 75 90 60 80 16
tdwg-tag mailing list tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag