Re: URL +1, LSID -1

Michel_Dumontier wrote:
> Unfortunately, this again demonstrates the problem in which the
> identifier for a biological entity - say mitochondrial Aspartate
> aminotransferase resolves to a nicely formatted HTML page. What if I
> have a semantic web application in which I would like to retrieve more
> information about this resource? Since the document is not an RDF
> document with machine understandable statements about it, it seems that
> my application wouldn't be able to learn anything more about
> http://purl.uniprot.org/uniprot/P12345

Yes, all your application has to do is follow the link-rel=alternate in the 
header of the web page, or set an Accept: application/rdf+xml header. This 
seems simpler (and more generic) than having to deal with an LSID resolver!


> Moreover, if I, a biological scientist, wish to make a statement (add
> new knowledge) about this protein, I certainly wouldn't want to do so
> using this identifier.

As a biologist I'd think that you would prefer to make statements about 
http://purl.uniprot.org/uniprot/P12345, because other, less enlightened 
biologists can enter this into their web browser and get something useful.


> IMHO, if you are going to use URLs to identify RDF resources

http://purl.uniprot.org/uniprot/P12345 does not identify an RDF resource, 
it represents our concept of some protein. There just happens to be an RDF 
representation at http://beta.uniprot.org/uniprot/P12345.rdf. But most of 
the concepts we use do not (anytime soon...) have such a representation.


> It's great to make human readable pages - I for one love nicely
> formatted pages. But please add a statement (say using the predicate
> HTMLPage or something to that effect) to an RDF document that the web
> page is located at http://purl.uniprot.org/uniprot/P12345.html (yai!
> which resolves correctly)

In the parallel universe where all the resources we reference are available 
in RDF, and only a few have web pages, that's the approach I chose :-)


> The use of a location free identifier such as an LSID provides me with
> the capability to make statements about resources that I care about.

http://purl.uniprot.org/uniprot/P12345 is as location free as an LSID!


> LSIDs and URLs can live together just fine. Using owl:sameAs predicate
> to bind them together is one easy way of doing this.  Just make sure
> you're talking about the same thing.

Entia non multiplicanda praeter necessitatem...

Received on Wednesday, 11 July 2007 07:17:18 UTC