RE: Cross-ontologies reasoning

At 13:50 -0800 12/19/03, Ugo Corda wrote:
>That's also the way I understood "humans in the loop". But my point 
>is that, even if you only need to set the mapping once, there might 
>be thousands and thousands of mappings needed because of the 
>potential proliferation of different ontologies within the Semantic 
>Web. So my concern is finding the skilled manpower capable of 
>handling that volume of work.
>
>Ugo

why does it have to be skilled?  there's an awful lot that people can 
do with good tools that will help make mappings without them even 
knowing it -- for example, if I say my pet is both a cyc:dog and a 
petshop:canine during some fill in the forms session (marking up 
pictures of my dog), som spider can come along later and bring that 
information along for other tools and or people to use.  I had some 
descriptions of a couple of scenarios of this a couple years back in 
the paper http://www.cs.umd.edu/~hendler/AgentWeb.html, and we've 
started exploring a number of tools in this space in my research lab 
...

>
>>  -----Original Message-----
>>  From: Danny Ayers [mailto:danny666@virgilio.it]
>>  Sent: Friday, December 19, 2003 1:36 PM
>>  To: Ugo Corda; public-sws-ig@w3.org
>>  Subject: RE: Cross-ontologies reasoning
>>
>>
>>  At 13:54 -0500 12/17/03, Jim Hendler wrote:
>>  > all the useful tools I know of are human in the loop and
>>  partial-mapping
>>  based
>>  [...]
>>  > cross ontology reasoning is the thing that makes Semantic
>>  Web different
>>  from all
>>  > other AI ontology work to date
>>
>>  These statements make a lot of sense to me. At the same time,
>>  they do not
>>  sound very encouraging as far as the Semantic Web is concerned.
>>  Unless the Semantic Web limits itself to a fairly small set
>>  of "canonical"
>>  ontologies (which is highly unlikely, given the open-ended
>>  nature of the Web
>>  itself), then the need of having humans in the loop seems to
>>  indicate that
>>  cross ontology reasoning within the Semantic Web is a very
>>  impractical goal
>>  right from the start.
>>  ---
>>
>>  Hmm, there are different ways you can read "humans in the
>>  loop". It may only
>>  be that they are needed once, to set up the mapping, after which the
>>  machines can be left to their own devices (so to speak).
>>
>>  Cheers,
>>  Danny.
>>
>>

-- 
Professor James Hendler			  http://www.cs.umd.edu/users/hendler
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  240-277-3388 (Cell)

Received on Wednesday, 24 December 2003 13:56:06 UTC