Re: RDF Issue rdfms-literalsubjects

>From: "Pat Hayes" <>
>>  >I don't understand this "damage" to syntax.  What are the problems with
>  > >using something like "rdf:aboutLiteral" instead of "rdf:about" where the
>>  >subject is a literal?   I gather that there is no problems with the
>>  >semantics of using literals for subjects.
>>  There are some syntax issues. One problem arises as follows. Imagine
>>  that we allow literals as subjects, and we have the following kind of
>>  situation
>>  [1] aaa --ppp---> "13" ---qqq---> bbb
>  >
>  > [2] ccc --rrr-->"13" ---sss--> ddd
>>  Now if we have 'tidy' literals - which many people want to, for
>>  technical reasons - then those two literals have to be the same node;
>>  and then we have a graph with five nodes consisting of a literal with
>>  two triples coming 'into' it and two triples coming 'out of' it,
>True .. see graph:
>>but no way to tell which goes with what.
>This seems to me to be a feature and not a bug ...  I guess I don't
>understand your concern.

OK, I didn't express it very well.

Some people want tidy literal nodes. Some people want to allow 
'multiple' uses of literals. As long as literals are not subjects, we 
can kind of finesse this mismatch of expectations by making the same 
literal 'contribute' different meanings to each triple it happens to 
be in.  But if literals can be subjects (as well as objects) then 
this will break, for the reasons given above. It is rather an arcane 
point, and I probably shouldn't have mentioned it.

If we have tidy literal nodes and fixed interpretations of literals 
(eg as strings) then there are no syntactic problems in graphs or 
Ntriples, I agree. But then....


>Well I can see a lot of documentation that would need to be changed .. but I
>can't see any technical problems with just doing it ... can you?  On the
>other hand I can see a lot of other things that could start happening very
>simply.  For example, we could just start talking about concepts like "The
>Coherence Theory of Truth" or "Pat Hayes"  without having to make a URIs for

Well no, because all you would have is the ability to talk about 
those character strings, not about what they might refer to. That's 
the cost of having tidy-literal graphs: it kind of forces you to lock 
down a single fixed global meaning for each literal.


IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax

Received on Sunday, 24 February 2002 23:00:18 UTC