Re: I have a trouble with The RDF Model

Some thoughts on Modeling ::

> 
> >This sentence has five words.
> >This sentence is in english.
> >This sentence begins with "This".
> >This sentence talks about itself.
> >
> >Look at the RDF version of the last example above :
> >
> >  [stidD, stidD, rdf:subject, stidD]
> >
> >Funny, isn't it ?
> >Not very useful, but definitely consistent.
> 

Modeling techniques can help.  Suppose we have
cycles as follows :

    [sid1, sid2, property1, object1]
    [sid2, sid1, property2, object2]

Then [sid3, sid1, is_reified, sid2] and
     [sid4, sid2, is_reified, sid1] are both true.

We can replace both sentences by a single 
sentence with a compound property to remove cycles ::

     [sid, common_subject, (property1,property2),
                           (object1,object2)]

In fact, 'is_reified' relationship between sentences
can be seen as a functional dependency between two
properties. If we represent a functional dependency
by an arrow (->) then we can formally define a model
as having a set of properties and a set of dependencies.
      a model M == {(P, Q, ...),
                    (P->Q, Q->R,...)}

A statement is a (property, value) pair (resulting
from a model)with a subject and sid attached.
I have a perception that all rules and axioms
of functional and multi-valued dependencies will
equally apply in RDF schema design.

A property is identified by M.P where M is the
model (or namespace) and P is the property name.
When we travel over the web from M.P to M.Q using
URIs, we have to make sure that the model M (used)
is the common model with a common design for
consistency.

--ssarkar@ayushnet.com

Received on Thursday, 30 November 2000 10:15:23 UTC