Re: Latest changes introducing the RDF API

Ivan Herman wrote:
> On Apr 18, 2011, at 14:10 , Nathan wrote:
> 
> <snip/>
>> For the RDF API, we've already discussed this previously where you yourself mentioned that often literals as subjects are needed for advanced usage, when querying and for rules etc, these modules wouldn't be able to use the API, or would have to break it compatibility wise, if we preclude literal subjects, so why go there just for the sake of consistency between specifications on an area that is already widely inconsistent and when the feature is actually required to process and use RDF in some cases?
>>
> 
> Actually, the issue for OWL RL implementation is with bnodes as properties rather than literals as subjects... But yes, you have a point there, I admit. It is not the issue of not being able to do that, but rather that it requires workarounds. I am still _very_ uneasy.

Ah yes, I should have checked my references on that one :p If it helps 
any, I can see no good / practical reason why that second level of API 
(like the RDFa API methods we mentioned earlier) couldn't have the RDF 
limitations, infact practically, I think it may have to.

> B.t.w., while we are at it: I do not know if you followed the f2f discussion on 'standard' skolemization. It may be a good idea to have such a skolemization built into the API, so that users would not be forced to implement that again and again... Using that, the bnode as predicate problem might disappear, because a user could use that in its own deduction process.

I caught the logs and tail end of it + followed on semweb list for a 
while, thought the same thing myself, and also that it could allow a 
GroundGraph type interface which had basic set operations defined on 
it.. which could be rather useful.

Guess we need to see if a standardized way of generating skolem 
constants is provided by the RDF WG, if it is, I see no good reason why 
we couldn't include it in the API.

Best,

Nathan

Received on Monday, 18 April 2011 12:37:35 UTC