- From: Knud Hinnerk Möller <knud.moeller@deri.org>
- Date: Tue, 14 Oct 2008 15:42:37 +0100
- To: Thomas Loertsch <loertsch.thomas@guj.de>
- Cc: Semantic Web <semantic-web@w3.org>
Hi Thomas, On 14.10.2008, at 15:33, Thomas Loertsch wrote: > OTOH standardizing eg a vocabulary for recipes alone is hard enough. > There's > no realistic hope for *the one* vocabulary to connect them all. The > semantic > web won't happen without some inferencing - be it through OWL or > some other > language. So, instead of developing a "vocabulary starter pack" > which could > at best be another DC, wouldn't it make more sense to establish a > clearinghouse/registry where mappings are published? Oh, I didn't mean a starter pack with new terms! That would indeed be a rather hopeless endeavour. What I meant was picking the most commonly used or most useful terms from well-known ontologies. This would amount to something like a best practices document. E.g., "If you want to express that something has a name/title/label, use rdfs:label." > That way a query for everything with a 'title' could automagically > sparql > the clearinghouse for every owl:equivalentClass of 'title' and add > those to > the original query. That's a scalable solution and it doesn't need > much > (actually none if I'm not mistaken) inferencing power either. That's an interesting idea. It would be interesting to find how this compares performance-wise to "ordinary" inferencing. Knud > Cheers, > thomas > > > On 14.10.08 15:43, "Knud Hinnerk Möller" <knud.moeller@deri.org> > wrote: > >> >> Hi, >> >> I'm all for reusing existing terms, see below. >> >> On 14.10.2008, at 10:48, Thomas Loertsch wrote: >> >>> Hi there, >>> >>> I want to derive a RDFa format from a microformat, an experimental >>> version >>> of hRecipe in this case. I think of them both as serializations of a >>> vocabulary. To make the distiction clearer I'll call them vRecipe >>> (the >>> implementation-neutral voacbulary), hRecipe (the microformat) and >>> aRecipe >>> (the RDFa format). >>> >>> I'm unsure how to map the vRecipe to RDFa. Should I reuse existing >>> vocabularies or should I develop a new one and then provide an OWL >>> mapping >>> to existing vocabularies? >>> >>> E.g. there's a title-element in vRecipe, called 'recipe-title'. In >>> hRecipe, >>> the microformat serialization, it's called 'recipe-title' as well. >>> Since >>> RDFa provides namespacing mechanisms it could be called >>> 'hRecipe:title' >>> here. A mapping from 'hRecipe:title' to 'DC:title' - >>> >>> hRecipe:title owl:equivalentClass dc:title >> >> During VoCamp Oxford last month a number of people including me >> discussed the problem that in the SW world there are too many >> different terms (properties and classes) that mean the same thing (or >> almost the same thing). Your title property is probably the best >> example: almost every new ontology or vocabulary mints their own new >> URI for saying "this is the name of that". There is rdfs:label, >> dc:title, foaf:name, sioc:title, doap:title, skos:prefLabel, ical:summary >> , ... A lot of ontology designers then go about and say that their >> new >> name/title/label property is a subproperty of rdfs:label, or >> equivalent to it. But as you say yourself, these assertions will only >> help if the right kind of reasoning is performed - and this, I'm >> afraid, mostly doesn't happen. The fact that RDF tools can parse the >> data doesn't solve the problem, because after that you'll also want >> to >> _do_ something witht the data. E.g., as a result, it is quite hard to >> query an RDF store for something basic such as "list me all the >> resources you have and their name". If there is no reasoning taking >> place, then all you can do is to include all possible label >> properties >> in the query, which is obviously not feasible. >> >> So, long story short: I think everyone should try to use the same >> properties and classes all the time, if possible, and as long as >> reasoning cannot be taken for granted. Maybe during the next VoCamp >> in >> Galway, I'm hoping to put together something like a "vocabulary >> starter pack" for the SW, suggesting a property or class for some >> basic annotation needs such "this is the name of that". I'm not sure >> yet how good the idea is, but I have a feeling something like that >> would be beneficial. >> >> Cheers, >> Knud >> >>> - would make it clear that both are semantically equivalent. >>> Although in >>> this example it's not obvious why not to use a straightforward >>> mapping to >>> the wellknown DC:title in the first place, but the whole vRecipe >>> vocabulary >>> needs mappings to a whole bunch of other vocabularies, some of them >>> not so >>> well known, and it looks quite messy when mapped straightforwardly. >>> It would >>> surely look much prettier - and was much easier to comprehend and >>> use - if >>> it was developed from scratch (and from the vRecipe voacbulary >>> respectively) >>> in a coherent way and *then* mapped to other, already existing >>> vocabularies >>> with owl:equivalentClass. >>> >>> I can see that the use of OWL adds complexity and that OWL can't be >>> handled >>> "meaningfully" by simple RDF tools but I'm not sure how much of a >>> problem >>> that is. Simple RDF tools can surely parse it which would be enough >>> for a >>> lot of usage scenarios. The whole power of the semantic web otoh >>> only comes >>> with RDFS and OWL and therefor it seems okay to me to use them like >>> I did >>> above. Or am I adding complexity where I really shouldn't? >>> >>> Cheers, >>> Thomas >>> >>> >>> .. >>> Thomas Lörtsch >>> Living at Home Multi Media GmbH >>> Redaktion Online >>> ... >>> Stubbenhuk 5 >>> 20459 Hamburg >>> .... >>> eMail: loertsch.thomas@guj.de >>> >>> >>> >>> >> >> ------------------------------------------------- >> Knud Möller, MA >> +353 - 91 - 495086 >> Smile Group: http://smile.deri.ie >> Digital Enterprise Research Institute >> National University of Ireland, Galway >> Institiúid Taighde na Fiontraíochta Digití >> Ollscoil na hÉireann, Gaillimh >> >> > > > > . > Thomas Lörtsch > Living at Home Multi Media GmbH > Redaktion Online > .. > Stubbenhuk 5 > 20459 Hamburg > ... > eMail: loertsch.thomas@guj.de > > ------------------------------------------------- Knud Möller, MA +353 - 91 - 495086 Smile Group: http://smile.deri.ie Digital Enterprise Research Institute National University of Ireland, Galway Institiúid Taighde na Fiontraíochta Digití Ollscoil na hÉireann, Gaillimh
Received on Tuesday, 14 October 2008 14:43:21 UTC