Hmm. I'm not sure if I'm adding value, but I'll try once more. (01)
At 12:26 PM -0600 1/24/08, Pat Hayes wrote:
>At 6:09 PM +0100 1/24/08, John Graybeal wrote:
>>I'd say it is one but we want ours to have more features and more consistent
>features, for example:
>>- access to term-level information
>
>That can be provided by a suitable interface. Existing SWeb editors such as
>Protege and COE do this. (02)
Well, yes, if you download the whole ontology and then ask for a particular
term. I'd kinda like to just ask for a particular term and only have that term
downloaded. (03)
>>- change tracking at file and term level
>
>Im not sure what this means ('tracking'?) (04)
Umm, you know, like, 'tracking'. Like revision tracking. Like change control
tracking. Like, keeping track. I seem to lack the right
(non-self-referential) words. How about: Telling me what I had before and what
I have now and what the difference is. (05)
And I'm sure Protege and others can difference two ontologies that I've
downloaded off the web. What they can't do is tell what what the state of
ontology X was last week, or last month. A controlled repository can do that.
And if it's a particularly nice one, it can do it at the whole ontology level,
or at the single term level. This is a service that is not provideable via a
straight web service. (06)
>
>>- access to all ontologies via consistent web service interfaces
>
>Surely this is exactly what the Web provides. (07)
Sure, if the only service you want is URL access. Multiple types of interfaces,
like give me this ontology in some other format, or give me the terms that
match certain patterns. (08)
>
>>
>>And many other features no doubt. :->
>
>But could they be best created as a collection of Web services accessing
>ontologies on the SWeb? (09)
Ah, well, I'd say no, because it introduces a level of indirection, and it
means downloading big ontologies and then processing them from beginning to
end, instead of (potentially) indexing to a particular term on the local
machine. (010)
But sure, it would work for most features, and it's an interesting model that
has a number of advantages. (011)
At this point I think we have to toss the thread over to the requirements
thread, which is a better home for the types of things I've been describing
anyway. I suspect that both implementation approaches can be used for many of
the requirements. (012)
John (013)
--
----------
John Graybeal <mailto:graybeal@xxxxxxxxx> -- 831-775-1956
Monterey Bay Aquarium Research Institute
Marine Metadata Initiative: http://marinemetadata.org || Shore Side Data
System: http://www.mbari.org/ssds (014)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/oor-forum/
Subscribe: mailto:oor-forum-join@xxxxxxxxxxxxxxxx
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/oor-forum/
Shared Files: http://ontolog.cim3.net/file/work/OOR/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository (015)
|