Re: [Treebase-devel] treebase.domain.name

[William P. <wil...@ya...>]

On Jan 14, 2010, at 11:57 AM, Hilmar Lapp wrote:
> Looks nice to me, but anyone (i.e., any third party) could do that on  
> top of each resource's individual base PhyloWS URL, couldn't they?  
> I.e., this need not be a responsibility of the data resources per se,  
> no?


Yes, although we don't want to proliferate synonymous references, right? 

On Jan 13, 2010, at 4:52 PM, Vladimir Gapeyev wrote:
> As far as I understand the problem, the real stability issues of a URL like http://purl.org/phylows/study/TB2:S1020 lie not in its "http://purl.org/phylows" part but in its "study/TB2:S1020" part


I take it that the "gold standard" is now the "linked list" approach in the semantic world: i.e., each RDF, XML, OWL, or whatever, uses fully resolvable and globally unique identifiers.  So for TreeBASE study_id "44", they would not use <id>44</id>, nor would they use <id>study/TB2:S1020</id> or even <id>/phylows/study/TB2:S1020</id> -- but rather they would use <id>http://purl.org/phylo/treebase/phylows/study/TB2:S1020</id>.

If there exists one, say, GIS document that references a tree like so: <id>http://purl.org/phylo/treebase/phylows/tree/T1020</id> and another, say, CITES document that references the very same tree with <id>http://treebase.org/phylows/tree/T1020</id>, a semantic reasoner might not realize that the referenced trees are the same tree, no?

So, my take of the "linked list" approach is that the full URLs matter (not just the "study/TB2:S1020" part), and therefore we should beware of multiple domains that have the effect of creating synonymous URIs. 

bp