RE: Mapping between different knowledge sources

Hi Kevin

> Mapping data requires that someone understand both data sources.
> Until machines are able to understand the data that they manipulate
> I would never trust them to generate conversions without the 
> prerequisite thought.
> 
> I think the best you can hope for is to provide good tools to
> help in the extraction or creation of data relevant to each
> user.  In that sense of 'interoperate between multiple vocabularies',
> I think we have decades of historical use to fall back on.  (And
> millenia of human experience.)
> 
> In that case RDF is meant to be a better tool for building those
> mappings, primarily because it represents relationships very 
> strongly.  But as you can see from the difference in representation
> of RDF's two basic statement types, the standard triple statement,
> and the four triples that constitute a reified statement none of
> which looks like the standard triple, mapping in RDF cannot be
> considered automatic IMHO.

Yes. I think we are in violent agreement here. Some of the semantic web
marchitecture (e.g. the use of terms like semantic interoperability) has led
people to conclude that RDF is some kind of magic bullet here. I don't
believe it is. To follow John Searle, at the end of the day, the computer is
just dealing with symbols so we need an observer in the system to match
those symbols up. This is why I think calling it the "semantic" web is
rather unhelpful. 

However even with an observer, there are some ontologies that are easy to
map and others that are difficult or impossible because (to use the jargon)
they have different "ontological commitments". The reason I posted the quote
is because the authors have a classification of how these clashes can occur,
and that seemed quite useful.

br, 

Dr Mark H. Butler
Research Scientist                HP Labs Bristol
mark-h_butler@hp.com
Internet: http://www-uk.hpl.hp.com/people/marbut/

Received on Monday, 24 March 2003 11:40:03 UTC