- From: Miles, AJ (Alistair) <A.J.Miles@rl.ac.uk>
- Date: Thu, 6 Nov 2003 14:12:01 -0000
- To: "'NJ Rogers, Learning and Research Technology'" <Nikki.Rogers@bristol.ac.uk>
- Cc: "'public-esw-thes@w3.org'" <public-esw-thes@w3.org>
Hi Nikki, I completely agree. So now I think nodes types as 'soks:Concept' should NOT be allowed to have a language property. ONLY the labels should be allowed to have a language property/tag. Representing non-exact equivalence relationships between concepts derived from different languages is then covered by whatever we end up doing for inter-thesaurus concept mapping. This does have consequences for constraining the data model. It means a node typed as a 'soks:Concept' must then be allowed to have multiple 'soks:descriptor' properties, one for each language. Is it then possible in OWL to express the constraint that a concept may have one and only one 'soks:descriptor' property for each language? If not, can we live with this? Al. -----Original Message----- From: NJ Rogers, Learning and Research Technology [mailto:Nikki.Rogers@bristol.ac.uk] Sent: 05 November 2003 12:32 To: Miles, AJ (Alistair); 'public-esw-thes@w3.org' Subject: Re: Design Issue (4) - Concepts as language-embedded, language indepe ndent, or both? Hi Alistair My philosophical angle on this is: Human society distinguishes itself from (the rest of) animal society by its reification of abstract concepts, done primarily via language. So the concepts don't come from the language, they exist independently - languages evolve in the attempt to articulate human understanding of abstract concepts. Different languages have evolved over time of course, and have much in common, but also some languages have sought to reify concepts that other languages don't, and so on... Coming back to the question of a pragmatic solution to modelling multilingual data: I tend towards Option 2, but then I've not experimented with much 'real' data, so you might want to give me some illustrative counter examples. Is it not the case that the *concepts* being reified by one language can be mapped - partially or otherwise - to concepts being reified in another language? And if there are "missing" concepts for some languages then that's ok? i.e. is it possible to always map thesauri at the conceptual level as opposed to at the 'label' level? I know this means deriving concept hierarchies from thesauri from thesauri that don't have this explicit - but I thought that was what we were aiming for anyway? I guess we should continue the discussion on the wiki ... Nikki --On 31 October 2003 17:31 +0000 "Miles, AJ (Alistair) " <A.J.Miles@rl.ac.uk> wrote: > > I've added this issue to the discussion on the RDF Thesaurus wiki page > <http://esw.w3.org/topic/RdfThesaurus> > > Here is a summary:- > > Issue 4 - Concepts as language-embedded, language independent, or both? > > There are multilingual thesauri. When modelling multilingual data in RDF, > we can choose one of the following options: > > Option 1: Concepts in a language - allow language properties only on nodes > typed as Concepts. > > Option 2: Labels in a language - allow language properties (or tags) only > on nodes (or literals) which represent labels. > > Option 3: And/or - allow concepts and/or labels to have language > properties. > > > The choice of solution has bold implications. If we choose option 1 we are > assuming that all abstract concepts are embedded in a language; there can > be no language independent concepts. If we choose option 2 we can model > only concepts that are deemed to be 'language-independent'. If we choose > option 3 we can represent both language-embedded and language-independent > concepts, however there may be some considerable scope for confusion. > > ... more at <http://esw.w3.org/topic/RdfThesaurus> > > > CCLRC - Rutherford Appleton Laboratory > Building R1 Room 1.60 > Fermi Avenue > Chilton > Didcot > Oxfordshire OX11 0QX > United Kingdom > > Email: a.j.miles@rl.ac.uk > Telephone: +44 (0)1235 445440 > > > ---------------------- NJ Rogers, Technical Researcher (Semantic Web Applications Developer) Institute for Learning and Research Technology (ILRT) Email:nikki.rogers@bristol.ac.uk Tel: +44(0)117 9287096 (Direct) Tel: +44(0)117 9287193 (Office)
Received on Thursday, 6 November 2003 09:12:06 UTC