Re: [WN] comments on draft

Jacco van Ossenbruggen wrote:

> Review of http://www.cs.vu.nl/~mark/wn/wn-conversion.html

In addition to my previous review, one more remark.

Under open issues, the document now states:
"One of the issues is whether or not to make the inverse properties 
"visible" to RDFS tools."

Following an off-list discussion between Mark and Jan Wielemaker, I only 
now understand that this line
refers to the fact that all reverse triples have been explicitly added 
"to help" RDF-only applications that do
not understand the owl inverse predicate.  From what I understand from 
Jan, this results in more than
a million extra triples (!).

I agree with Jan that these extra triples should be not part of the core 
translation:
For OWL users,  over 1M of redundant triples is very inconvenient, 
rule-based systems have probably more efficient
ways to deal with this problem and even many RDF users would probably be 
happy to formulate their query in
both directions explicitly if this reduces their triple store with more 
than a million triples. 

Another argument against explicitly added inverse triples is that if 
inverse triples are added explicitly,
the same should be done with transitive triples, which is not the case 
either.

Of course, I have no objection against providing these triples in a 
separate file as an extra service.

Thanks to Jan for bringing this up,

Jacco

Received on Monday, 28 November 2005 14:41:09 UTC