Fwd: Current state of the discussion around RDF+OWL & Topic Maps?

Apologies if everyone gets this twice.  Apparently, email is a tad beyond my grasp today...

Begin forwarded message:

> From: "Andrew S. Townley" <[email protected]>
> 
> Hi Lars,
> 
> On 11 Nov 2010, at 12:47 AM, Lars Heuer wrote:
>> 
>>> hoping that someone could point me to the most recent reference
>>> point in this comparison.
>> 
>> I am not aware of recent efforts. RTM seems to be a complete mapping
>> but it does not solve all problems. It's just a simple mapping where
>> you can state that the RDF property X should map to the Topic Maps
>> construct Y.
>> 
>> RTM is practical: You can map the RDF predicate X to occurrence Y and
>> state that Z should be treated as topic name A but not more. You
>> cannot interpret xml:lang as a theme (CRTM does, btw ;)).
> 
> Right.  Good to know (shameless plugs aside... ;)
> 
>> 
>> I've had the "fun" to map DBpedia (RDF sources) to some extent to
>> Topic Maps and from this experience I'd say that RTM solves many
>> problems but does not cover all possible problems (and I don't think
>> that the current RTM definition promises that all your RDF -> Topic
>> Maps problems would be solved).
>> 
>> RTM is good, but you need some context and RTM should conditions like
>> "if X is an xsd:URI treat it as an occurrence, otherwise as a name".
>> Neither RTM nor CRTM does provide conditions.
> 
> Also good to know.  I knew you'd done some things with DBpedia for sure, and there are a few other sources out there that I was investigating.  My concerns were around expressiveness and semantic mismatches in underlying assumptions between the two, but the more practical and pragmatic observations from experience are also invaluable.
> 
>> 
>> An RDF to Topic Maps mapping should be (or must be) context-sensitive.
>> It's not a bug in Topic Maps but a failure in RDF ... or a feature in
>> Topic Maps.
> 
> 
> This is about where I was with it.  I've seen stuff about adding context through quads to RDF, but for the most part, it kinda seems to ignore the issue.  Looking at some of the points raised in the NetworkedPlanet article on interpreting RDF, I hadn't really realized that RDF was as big into reification.  It seems very clunky, so I'm not surprised that the opendata/linkeddata folks say to avoid it.
> 
> Maybe I'm not understanding things correctly, but given its inferencing bias, it seems like most RDF+OWL is intended to be expressed as absolute statements within an implied context, but which are actually shared on a global level.  I'm sure you can, but it doesn't seem like it's geared as much towards expressing context for the information as a whole vs. the individual assertion as is possible with the Topic Maps specs.
> 
> I guess more specifically, I'm interested in knowing the bounds of how some of the identity and context expressions possible in Topic Maps are or are not possible using RDF+OWL as it seems to be used in the wild and at scale.  It seemed to me that our community would be more aware of these issues than the RDF folks (certainly of the ones I've ever talked to).
> 
> Thanks for the info.
> 
> Cheers,
> 
> ast
> --
> Andrew S. Townley <[email protected]>
> http://atownley.org
> 

--
Andrew S. Townley, Managing Director
------------------------------------
Archistry Limited
3 Lombard Street East, Suite 115, Dublin 2, Ireland
Phone:    +353 1 293 2998 * Fax: +353 1 293 2999 * Mobile: +353 86 996 2490
Email:    ast-MrgM/8S5P0jqlBn2x/[email protected] * http://www.archistry.com
Twitter:  http://twitter.com/atownley
Inspiring business innovation (TM)

*******************************************************************************************
Archistry Limited is registered in Ireland at 168 Corbawn Wood, Shankill, Co. Dublin.
Company number IE 422089; VAT number IE 9572111G; Directors A. Townley (US), S. Carraz (FR)
*******************************************************************************************

--
Andrew S. Townley <[email protected]>
http://atownley.org