Re: [Fenfire-dev] PEG swamp_easier--benja: An easier API for Swamp

[Tuomas Lukka]

[sorry for the long quote, seems necessary]

On Mon, Sep 22, 2003 at 02:51:30PM +0300, Benja Fallenstein wrote:
> Tuomas Lukka wrote:
> >On Mon, Sep 22, 2003 at 02:29:49PM +0300, Benja Fallenstein wrote:
> >
> >>Tuomas Lukka wrote:
> >>
> >>>Why *should* it work for "equivalent" triples?
> >>
> >>Because the same thing can have more than one name (node in the graph), 
> >>and if e.g. "c = d," then the triple "a b c" entails "a b d." Not 
> >>allowing "a b d" to be in a graph at the same time as "a b c" is like a 
> >>calculator that can compute 7+3+2, but not 7+3+0+2.
> >
> >As long as the definition of equivalence is as fuzzy as you have made
> >out, 
> 
> I haven't made it fuzzy at all. There is a set of resources; there is a 
> set of names for these resources, and a resource can have more than one 
> name (or no name at all). Two triples of *names* of resources (i.e., 
> nodes) are equivalent if the subjects name the same resource, the 
> predicates name the same resource, and the objects name the same 
> resource. If you see any fuzziness there, I suggest you get a little 
> clearer.
> 
> Also, I'm not making any definitions, I'm just trying to explain. If you 
> want to have a look at the sources, see for example 
> http://www.w3.org/TR/owl-ref/#IndividualIdentity .
> 
> >I think it's fine: I see it more like 7+3+2 != 7+3+2+epsilon  
> 
> I don't see any relation between this and the issue at hand.
> 
> >If someone wants equivalences to work, what they can do is *unify* 
> >the equivalent nodes of the graph **before** running code on it.
> 
> Knowing which names represent the same resource is not generally 
> possible. 

*THIS* is the fuzzy part. I'm fine with having resources have different
names, but this is the part I don't like.

> You can know *some* equivalences, but not all. (E.g. if you 
> have a property with cardinality 1, but more than one triple with that 
> property and the same subject, then you know that all the objects are 
> names for the same resource (or the graph states a contradiction). This 
> is why returning an arbitrary one of the objects in ``getObject()`` 
> should be fine-- they should all represent the same resource.)

Same here: why not run this *before* running code, cleaning up the graph.

For example, the graph could have been merged from two graphs
where different values are given.

I understand that what you're saying is the equivalence idea of RDF
in general, but I still think it's not appropriate for Swamp; things
like this should be preprocessed out to get a *clean* graph.

        Tuomas