Re: a few more use cases for completeness

At 9:17 AM -0500 12/7/01, Leo Obrst wrote:
>A question I have: under which use case is conceptual search? I too had
>posted this as a use case, but am not sure where it falls.
>
>Thanks,
>Leo


In creating my categories I had this one falling in 
collection/archives, but only because I figured in the short term any 
sort of conceptual search mechanism was likely to be restricted to a 
particular domain or collection -- but you are the folks making the 
decisions as to how to ground out the various use cases

Also, I should note that not every use case needs to fit neatly in 
one category.  These were collected to give us organizational 
structure and some examples - I don't think there's an intent to be 
complete in any respect, as that would likely be impossible.

If I take off my chair's hat of neutrality for a minute, I'd say that 
I see the following as critical differentiators

  1) Collection Mgt - can assume there is a fixed set of sources at 
some moment in time, that it is relatively finite (although possibly 
large) and probably domain restricted in some way. Focus is on the 
"metadata" helping to manage the documents/resources in the 
archive/collection - consistency/computability of some kind seems 
desirable.

  2) Interoperability - I assume here many heterogeneous sources, 
probably challenging any assumption of closure and/or decidability - 
the closure over the data descriptions could be all data on the web 
-- clearly too large to worry about issues like decidability (even 
polynomial is not good enough when your N is in the billions) and 
probably impossible to provide anything other than local 
consistencies.  We need to make sure we can live/compute with this

  3) Web Services - as well as the usual business case (i.e. people 
see $billions in this) the challenge of web services to ontologies 
seem to me to come in the fact that we need to describe/represent 
processes and activities - not the usual "nouns" what we think about 
in the class hierarchies of RDFS.  WIll the expressivity of DAML+OIL 
give us enough power to do this?  Are there other constructs we may 
need to consider informed by previous work in UML, in Agent 
Communication Languages (ACLs) and in the various FIPA use cases?

  4) WOL requirements - this group has a tough mission -- it needs to 
figure out a strong intersection of the needs above (not clear if the 
union would be doable, for example archive management might want to 
enforce global consistency while interop might require only some sort 
of local agreement).  However, it also needs to figure out if there 
are things that span these areas, critical to all, but not 
necessarily likely to arise in any particular one -- a good example 
Jeff raised is versioning, which would be an important language 
feature for all of the above.



-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
AV Williams Building, Univ of Maryland		  College Park, MD 20742
http://www.cs.umd.edu/users/hendler

Received on Friday, 7 December 2001 14:46:08 UTC