- From: chris mungall <cjm@fruitfly.org>
- Date: Fri, 12 May 2006 04:59:17 -0700
- To: Matthias Samwald <samwald@gmx.at>
- Cc: <public-semweb-lifesci@w3.org>
On May 11, 2006, at 4:18 PM, Matthias Samwald wrote: > >> Hmm, the 10^16 genes instantiated in the volume of space occupied >> by me are neither irrelevant (to me anyway), nor are they concepts. >> They are very real instances of physical material objects - at >> least under one definition of gene. > >> >> in your example I presume the ID gene/123 was intended to be an ID >> for a gene type rather than an instance - or perhaps not? > > It could also be a URI for an OWL class. The 10^16 genes (in the > 'physical object' sense) in your body would be instances of this > class - we would probably never make this instances, though. Using > classes would be the most consistent way to do it. exactly > > However, I am not an advocate of doing everything with OWL classes, > as it is hard to implement - actually I am very opposed to that > idea. I am preferring a two-world approach: unifying realist > descriptions of spatiotemporal particulars (e.g. one of your genes) > with conceptualist descriptions (e.g. the concept of 'human insulin > receptor gene'). In this model, the conceptual description are used > to annotate the realist descriptions. However, I will write more > about that some other time, as this is not the topic of this > discussion. I agree that OWL classes would present many implementation difficulties here. I'm interested in what the alternatives are, given that I think we both want to avoid unnecessary abstractions such as "instance of a database record". I'm wondering how your concept of a conceptual description differs from owl classes or from my types. > > Either way, aiming to describe mere concepts (and not the 'real > things' themselves) is still much better than simply describing > database entries and their relations to one another. If we just > want to have better interoperability between database entries, > generic XML with XLink would suffice - there would be no > significant need for RDF or OWL, in my opinion. I pretty much agree, with the caveat that I would include types as being real, and I'm more interested in types than concepts - but that's another discussion entirely cheers chris > > > > //Matthias > >
Received on Friday, 12 May 2006 12:04:21 UTC