Re: Xenon and Fresnel

From: Emmanuel Pietriga <>
Date: Mon, 06 Dec 2004 13:12:01 +0100

Chris Bizer wrote:

> He has developed an alternative approach to Fresnel called Xenon which
> is more XSLT oriented. A paper submitted to WWW2005 about Xenon is
> attached to this mail.

I read the www2005 Xenon paper last week, and I found it very
interesting. Some comments about the Fresnel/Xenon interaction below.

> We were discussing how Xenon and Fresnel could fit together (see mails
> below) and Dennis proposed, that Xenon could be used to implement
> Fresnel. Interesting direction ...
> What do you think?

As you said, Xenon is lower-level than Fresnel, and from what I read of
Xenon it seems indeed that this could work fine (with some adjustments).
Ideally, Fresnel could represent a way to get familiar with Xenon at a
high level, and users who need more expressive power could go deeper
into Xenon from there.

> What do you think about Xenon?

I like the XSLT-inspired template approach, but I think Xenon is much
harder to learn than XSLT. One reason for this is that it is RDF-based,
which makes its syntax and model more complex. So it seems powerful, I
find it elegant, but I'm not sure it can reach a lot of people "as is".
Maybe with some kind of front-end. Or higher-level specification means
such as a fresnel implementation on top of it (as a start).

> I was also talking about Fresnel with some guys who were presenting
> semantic portals at ISWC. They had similar, less expressive and more
> ad-hoc display languages in their portals and were very interested in
> Fresnel. It seams like many people are seeing the need for agreeing on a
> display language and we could get some adopters easily, if we were
> coming up with something flexible, but easy to use ....

That would be nice.

>>> I totally agree with the goals which your are formulating in section
>>> one,
>>> paragraph 5 and also like your abstract model in section 2. Especially,
>>> the idea of a "template marketplace" and the loose coupling between
>>> templates
>>> using roles. I think giving the user the possibility to choose between
>>> different representation is also very important and one of the key concepts
>>> of Fresnel.

So do I. Heterogeneous composition for presentation is very important in
the context of RDF.

>>> The interesting question is now: Which of our approaches fits which
>>> requirements?
>>> I thing your language (being functional) is more expressive than our
>>> declarative approach. But you are buying this flexibility with quite
>>> some
>>> complexity. My idea with Fresnel was to keep things as simple as
>>> possible
>> in
>>> order to make it easier for users to learn the language. Most people
>> easily
>>> understand HTML and CSS, but I know quite some which are having problems
>>> with XSLT.

Considering this, having a Xenon implementation of Fresnel would blur
the frontier between the two.

>>> As I already said at ISWC, we are using an RDFpath like selector
>>> language,
>>> where you are using SPARQL.
>>> ?TARGET/_at_vCard:TEL [
>> _at_rdf:type=<>]
>>> Most of your example queries aren't to complex and don't use the
>>> expressivity of SPARQL. So it might be an idea to switch to RDFpath in
>> order
>>> to achieve better readability.

I haven't looked thoroughly at SPARQL yet, but I am really much in favor
of a powerful-enough RDFPath language.


Emmanuel Pietriga
INRIA Futurs - Projet In Situ    tel : +33 1 69 15 34 66
Bat 490, Université Paris-Sud    fax : +33 1 69 15 65 86
91405 ORSAY Cedex  
Received on Mon Dec 06 2004 - 12:12:46 EST

This archive was generated by hypermail 2.3.0 : Thu Aug 09 2012 - 16:39:17 EDT