W3C home > Mailing lists > Public > semantic-web@w3.org > September 2009

Re: modelling issue?

From: Frank Manola <fmanola@acm.org>
Date: Sun, 27 Sep 2009 18:02:23 -0400
Message-id: <7F2281C9-30E1-4127-B3DD-6666AE70DF27@acm.org>
Cc: semantic-web at W3C <semantic-web@w3c.org>
To: paoladimaio10@googlemail.com

On Sep 27, 2009, at 4:58 PM, Paola Di Maio wrote:

> also I dont see how what you discuss resolves the problem of having  
> RDF used to represent different constructs
> class:relation:class
>  but also
> class:attribute:value
> but I admit I do not have the time (nor the competence) to work on  
> this myself, however from where I stand
> I see some potential conflict and unreliability -

I don't know if this helps or not, but it may be worthwhile to note  
that RDF essentially uses the same modeling approach that relational  
databases (and predicate logic) do, namely modeling both "attributes"  
and "relationships" as (mathematical) relations.  E.g., modeling  
something using only 2-ary relations (kind of like RDF), you would  
represent both "attributes" of Joe, like his weight, and  
"relationships" between Joe and other things (like Joe's brother) as  
tuples of simple relations like:

weight(Joe'sID, 25)
brother(Joe'sID, Fred'sID)

Note that the binary relations involved correspond to RDF properties,  
and the tuples correspond to RDF triples:  the names of the properties  
here are  "weight" and "brother" (except that RDF would use URIs for  
property names to distinguish them within the Web, as opposed to a  
relational database which only needs to distinguish relation names  
within a single database).  Both weight(person, number) and brother 
(person, person) are mathematically relations;  the distinction  
between "attribute" and "relationship" is a conventional one:  that  
it's an "attribute" if the relation is between an "entity" and a  
literal value, and it's a "relationship" if the relation is between  
one "entity" and another.  But neither RDF nor relational databases  
have this distinction between "attributes" and "relationships" "built- 
in" in any sense;  everything is represented by relations.

You state that you see some potential conflict and unreliability.   
Could you describe these a little more, since I'm thinking that that  
if these are problems with RDF, they are problems with relational  
databases too?   Or could you describe what you believe the difference  
is between modeling in RDF and modeling in relational databases (aside  
from RDF restricting you to binary relations)?

Received on Sunday, 27 September 2009 22:03:16 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:15 UTC