Re: DATATYPING: second draft

On 2002-01-24 18:31, "ext Pat Hayes" <phayes@ai.uwf.edu> wrote:

>> At 10:19 AM 12/14/01 -0800, Sergey Melnik wrote:
>>>> Dan has raised an issue, rephrased by Pat as how many triples result from
>>>>  merging:
>>>> 
>>>>     foo bar "10" .
>>>> 
>>>>  and
>>>> 
>>>>     foo bar "10" .
>> 
>> Ha!  I came up with exactly this issue in my review of the latest
>> model theory draft.  Do we have a consensus yet?
> 
> Well, let me suggest what the answer should be. This may seem odd, but....
> 
> I think the answer depends on whether this is a single graph being
> 'merged' with itself, or whether it is two distinct, but identical
> (ie isomorphic) graphs. If the former, then the 'merge' is just the
> same graph, with 2 nodes and one arc: a single triple. If the latter,
> then it is a larger graph with three nodes (assuming foo is a uriref)
> and two arcs. This is because different occurrences of a 'bare'
> literal have to be treated as distinct nodes (in case one of them
> should get itself attached to a different datatyping scheme from the
> other...), unlike urirefs. Notice that if you did a similar exercise
> with two copies of an all-uriref triple:
> 
> foo bar baz .
> 
> then the merged graph would be the same in either case, ie it would
> simply be a single triple; since in this case, the merging would
> (re-)identify the two copies of each uriref.
> 
> Pat
> 

Thanks, Pat, for that excellent clarification.

This is what is presumed by the TDL proposal, that literal labeled
nodes do not participate in tidying, and that every occurrence of
a literal has a unique node in the graph.

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Friday, 25 January 2002 05:59:28 UTC