Lists Home |
Date Index |
Only RDF/XML. RDF itself is quite nice. And powerful.
Now, the purpose in an experiment like "CSS syntax for RDF" is not
necessarily to produce something that can breeze through a CSS3 engine in a
browser, but to leverage a familiar syntax and developer experience.
The easier it is to produce metadata, the more it will get produced. The
more metadata there is, the better.
I'd like to expand this experiment, if I have time.
If you do a research on MCF or MCL you'll find something about the first
incarnation of the RDF syntax which were, by the way very similar to css.
Guha and me worked on that version. As you know an RDF object is a frame
(based on AI frames) and a frame contains properties. In those days (At
Apple research labs) we called each RDF object a unit. A unit would be
defined as follow:
Unit: (the unit identifier here)
Quite easy no? Moreover, an HTTP transaction could be treated as an RDF
frame so would be an SMTP transaction. Obviously the first line (an HTTP or
SMTP verb) could not be treated as an RDF frame member but all the other
name:value pairs could be. The main advantage we thought of at that moment
(I am talking here of very old times, something like 1996!! Gee we were
young and fool and... :-) was that a transaction (SMTP or HTTP) cold carry
meta information about the transported document. IN fact, in some ways, when
an HTTP header tells us that the contained document is of a particular MIME
type, this header is providing some meta information about the contained
When Guha was still working for Apple (before Netspace) he did a viewer
named Project X (the code name for this project at Apple Research). The
viewer let you see a constellation of information units as you can do today
with (but was less fancy than) The Brain applet. On my side I did a
Microsoft Window shell extension that allowed you to see a hierarchy of web
information units on your desktop.
I remember that before the XMLification of MCF (happening when Apple
research become history and Guha moved to Netscape) we were working on
associations between information units, something like the topic map
"associations". Implicitly, information units where hierarchical. For
instance, an information unit defined in a file would point to an other one
contained in another file. That way you could navigate from information unit
clusters to information unit clusters. We soon discovered from our zealot
supporters that new relations between topic would be a nice feature. During
the Project X experiment and since Apple zealots, like always, where very
supportive of this project, we got several site supporting the MCF format.
This allowed you to navigate from site to site or from topics to topics
inside the Project X viewer or the desktop Windows Explorer extension.
Now back to the present. With the RDF based RSS you get somewhat this
except that the RSS files are not linked together as where the MCF files.
With MCF I could publish my own mini web or my personal version of the web.
MCF was based on the notion of linkage (at least the latest version). This
is unfortunate that I do not see that kind of things now. RSS and most RDF
based system are still seeing the world as flatlands, we need a Christopher
Columbus to show us that flatlands are a subset of a rounded universe. Or
maybe I should say, since we already got the Columbus, that we need an
Americano Vespucci with good political skills to give a name to discovered
lands and status to previous discoveries :-)
Didier PH Martin
Note: for those not knowing what a frame is, think of it as a kind of free
form record. No needs to define a schema of what will be contained in the
record, just add properties/values about a particular object. The notion of
frame came out of two sources :
- lisp allowing free form lists
- knowledge bases simulating a real world view about a limited set of object
and their associated properties.