Re: Meaning of RDF:Statement

Dan Brickley wrote:
> 
> On Mon, 13 Dec 1999, Pierre-Antoine CHAMPIN wrote:
> 
> >   [Stating]  <--type-+[local-URI]--states--> [unique-URI]+-type--> [Statement]
> >   "12/13/99" <--date-+                                   +-subject--> [...]
> > "champin" <--author--'                                   +-object--> [...]
> >                                                          `-predicate--> [...]
> >
> > Such a dichotomty between the unique universal "statement"
> > and each local "stating" might be useful ;
> 
> Yes. The latter being more of an 'event', and perhaps sharing structure
> with other events (publications, agreements, properties of agents
> etc)...?


There should be a standard way to do this.

If statements is unique, there would have to be something like the
above.



This is obvious, but I would like to clarify anyway:

If there is only one URI for a specific statement, it could look like
this:

statement1 --attributedTo----> Person1
statement1 --attributedTime--> 19991101
statement1 --attributedTo----> Person2
statement1 --attributedTime--> 19991218

How do you tell which statement time belongs to which person?



There would have to be a URI for each statement 'event' if you want to
say more than one thing about a statement.


With models, it may be common to group some statements to a model and
say something about the statements as a group. Most things stated in
the real world takes more than one triple to model.

We have discussed this before. Should the model be modeled as a bag of
statements or should every statement have a direct pointer to the
model URI?



I would like to be able to optimise the statment storage in the
database. For every statement there could be a predicate, subject,
object and a model. That would make it more efficient in handling
diffrent models. And that would mean that there would be one URI for
each statement, even if they were identical. And that would let you
model the attributes about the statements in a more simple an
efficient way, compared by what would be required if the statement
were globaly unique, as described above by CHAMPIN. You would then
have:

statement1 --attributedTo----> Person1
statement1 --attributedTime--> 19991101
statement2 --attributedTo----> Person2
statement2 --attributedTime--> 19991218

... rather than the more elaborate version.

The equality of two statements could be determined by comparing their
p, s and o attributes.


The URI cold maby be generated by combining the triple with the model
or the origin. A combination with the model would be hard if the model
is generated by the triple URIs. A combination with the origin would
implicitly also bind the model to a specific origin.



With models; many attributions could go via the model:

statement1 --model--> model1 --attributedTo--> person1

We would still have to encapsulate the attribution:

statement1 --model--> model1 --states--> stateEvent --attributedTo-->
person1


But if a statement only has one model, it would be a bit faster.



That leads me to other optimizations. In my DB, every resource could
have the type as a part of the record. But only if there's never more
than one type per resource.

-- 
/ Jonas  -  http://paranormal.o.se/perl/proj/rdf/schema_editor/

Received on Saturday, 18 December 1999 08:51:08 UTC