Re: InverseFunctional properties are the new URI?

Internet Business Logic wrote:
> Simon --
> You wrote....
> "A whole class of interesting applications of the Semantic Web are 
> likely to arise where smushing is required without either URIs or IFPs."
> Do you have please an outline of such an example?
> Or, even better, some RDF data plus smushing requirements.  Something 
> like the task in [1].

A lot of people (I assume) are working on counter terrorism applications 
that smush partial descriptions from multiple databases and the web but 
that's not really what I was thinking of. I was referring to two more 
general problems:

1) merging information extracted by web/data mining from different sources

2) merging data from different vocabularies (i.e. the ontology alignment 
and mapping problem)

Arguably, 1 isn't mainstream Semantic Web (although I'd love it to be 
because that's my research interest and it would help us build a SW 
layer for the legacy web). However 2 is central to the SW and yes I know 
loads of useful work can be done without 2, but much more could be done 
with it.

Both these smushing problems involve uncertainty and even humans doing 
these tasks will often not agree on a definitive answer so Dan's right 
to sound a word of caution to prevent overselling of the SW along the 
lines of AI in the past.

Adrian - I know this isn't the concrete statement of a challenge that 
you were after. That would take a while to put together although trying 
to do that would be a useful exercise for me so (time permitting - I 
have to finish our foaf work first) I'll try and do that.



> It seems like a semantic "problem worthy of attack, that will prove its 
> worth by hitting back"  [2]
> Thanks in advance,      -- Adrian Walker
> [1] 
> [2] I think this is a quote from Knuth


Simon Price, Technical Consultant, Internet Development Group
Institute for Learning and Research Technology

Received on Friday, 30 July 2004 06:05:28 UTC