W3C home > Mailing lists > Public > public-swbp-wg@w3.org > June 2004

Re: [WNET] Draft summary of WNET proposal document

From: Danny Ayers <danny666@virgilio.it>
Date: Fri, 25 Jun 2004 22:43:32 +0200
Message-ID: <40DC8E74.3000009@virgilio.it>
To: "Aldo Gangemi (by way of \"Ralph R. Swick\" <swick@w3.org>)" <gangemi@loa-cnr.it>
Cc: public-swbp-wg@w3.org

Aldo Gangemi (by way of Ralph R. Swick <swick@w3.org>) wrote:

[Draft summary of WNET proposal   document]

Generally looks good to me, couple of thoughts -

The following paragraph seems a little out of place in the proposal, I 
wonder if it might be expressed as more concrete next-actions, i.e. to 
identify the choices and/or implement them.:

> <http://lists.w3.org/Archives/Public/www-archive/2004Jun/0019.html>Although 
> this first goal does not aim at an ontological interpretation of 
> WordNet (that is the second goal of the TF), it already presents some 
> choices, which are partly common to all attempts to port database 
> (schemas) to Semantic Web languages (e.g., should we create a place in 
> RDFS or OWL for each key in the source schema, even if its only use is 
> bound to database implementation?).
> Some other choices concern shallow interpretation of the WordNet's 
> structure, and can be found in the comments included in the 
> abovementioned datamodel.

Dropping into the specifics, regarding what should be mapped, I'd 
suggest there are two considerations - knowledge integrity and ease of 
implementation. Presumably we don't want the translation to RDFS/OWL to 
be lossy,  though it may well be reasonable to lose keys if they don't 
form part of the actual data. Conversely it may be useful to carry 
across 'struts', and even introduce new ones if it makes a significant 
difference to practical requirements.

Once a schema is in place ideally the translation of the instance data 
should be completely automated (determined by a declarative mapping). My 
guess is that this would be feasible, but with the expectation that 
manual cleanup will be needed in places due to quiet inconsistencies in 
the original data. Hopefully once such problems have been identified, it 
would be possible (and in everyone's interests) to feed required 
adjustments back to the original data.

Back to the proposal, a possible addition - it may be worth mentioning 
versioning issues.  It would be nice to get versioning support into both 
the schemas and the data early on, to help smooth any future transitions 
(i.e. to WordNet 2.1). Ok, I'm jumping the gun even more, but this may 
also help looking back to interop with WN 1.7 systems.



Received on Friday, 25 June 2004 16:45:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:09:39 UTC