RE: missing bit of RDF for XML people

Coming from an XML-ish background, I found another example quite useful in
illuminating the differences between RDF and XML.  I can't remember exactly
where this came from... but it's along these lines:

Imagine this XML:


If a person looks at it, you can clearly see the meaning.  Canada is a
country, Montreal is a city in Canada.
Now imagine this XML


What is the relationship between Toronto and Canada?  What exactly is Canada
according to this XML?  
Although the element name "country" is meaningful to a person, there's no
explicit meaning to a SAX parser (for example).

Don't know if this would help others, but I found it useful.


-----Original Message-----
From: Karl Dubost [] 
Sent: Thursday, February 03, 2005 4:30 AM
Subject: missing bit of RDF for XML people

Hash: SHA1


this is all naive but it's in my mind and writing the email helps me to 
push it out, and let me sleep :p

It's on the usual stance of XML People saying yes there is semantics in 


XML People: You can see that the hierarchical organization and the name 
of elements is the semantics.
RDF People: No, the semantics is not "machine explicit". (subject, 
predicate, object).

What RDF people are saying is that the predicate is not given, except 
in the specification which describes the semantics of each elements of 
the XML file.

So I thought let's write something without thinking about writing good 

  subject        predicate          object

country  - (blank_predicate1) --> "Canada"
country  - (blank_predicate2) --> city  - (blank_predicate3) --> 
country  - (blank_predicate2) --> city  - (blank_predicate3) --> 
country  - (blank_predicate2) --> city  - (blank_predicate3) --> 

A blank predicate would be something, there's a relation between these 
two things, but I don't yet how to describe it, just I know that it 
exists and I know that between these type of things, the relation is 
always the same unknown type.

Exactly like we say, there is this thing but I don't have a name for it 
(blank node)

Except that this is not possible in RDF, there's no such things as 
blank predicate. But maybe all of that is very naive, and doesn't make 
sense at all. :))))

[04:37] 	<karl>	question RDF qui me tracasse dans le lit :D
[04:37] 	<dom>	:)
[04:37] 	<karl>	sujet - (verb) -> blank_node - (verb2) -> something

[04:38] 	<karl>	mais
[04:38] 	<karl>	sujet - (blank_verb) -> something - (verb2) -> 
something1    OK
[04:38] 	<dom>	(on ne peut pas mettre de blank node comme predicate
[04:39] 	<karl>	est-ce qu'il est possible d'écrire un graphe dont on
connait pas les verbes
[04:39] 	<dom>	je crois que c'est hors du model RDF en général, pas

simplement RDF/XML d'ailleurs
[04:39] 	<dom>	par manque de use case, j'imagine
[04:39] 	<karl>	ahhh intéressant, je pense que c'est le probleme de
versus RDF. :)))
[04:40] 	<karl>	car XML c'est justement le second cas
[04:40] 	<karl>	on a des choses dont on connait la hiérarchie
[04:40] 	<karl>	mais dont on ne connait rien de la relation entre
[04:40] 	<dom>	hmm... je suis pas sûr que ce soit aussi simple ; ça

dépend de chaque vocabulaire XML autant que je puisse dire
[04:41] 	<karl>	oui bien sur
[04:41] 	<karl>	mais quand tu as
[04:41] 	<dom>	(ce qui est à mon avis au coeur de la différence XML
RDF/XML - RDF/XML lie une structure à une sémantique)
[04:41] 	<karl>	<pays><ville>montreal</ville></pays>
[04:41] 	<karl>	les tenants d'XML disent c'est semantique
[04:42] 	<karl>	et RDF dit non car c'est pas explicite
[04:42] 	<karl>	parce que la hierarchie
[04:42] 	<karl>	pays - (blank_something) -> ville
[04:42] 	<karl>	n'est donné que dans la spec
[04:42] 	<karl>	humainement
[04:43] 	<dom>	yup, tout à fait
[04:43] 	<karl>	pays - (blank_something) -> ville -
- -> "montreal"
[04:44] 	<karl>	c'est pour cela je me disais si RDF permettait de
cela :) comme les blank_node cela permettrait de commencer à joindre 
les deux bouts
[04:44] 	<karl>	dans le sens, ouaip il y a une hiérarchie mais on ne

sait pas qu'elle est la valeur semantique de cette hierarchie
[04:45] 	<karl>	on peut ecrire "incomplet" qui pourrait se completer

par le futur
[04:45] 	<karl>	* ecrire un graphe "incomplet"
[04:50] 	<karl>	Je retourne me coucher pour l'instant. :))) fallait
je l'écrive.
[04:59] 	<karl>	 dom
[04:59] 	<karl>	bon je vais me coucher

- -- 
Karl Dubost -
W3C Conformance Manager
*** Be Strict To Be Cool ***
Version: GnuPG v1.2.4 (Darwin)


Received on Friday, 4 February 2005 07:48:26 UTC