On Tue, Jun 17, 2008 at 8:10 AM, Sylvain Hellegouarch <[EMAIL PROTECTED]> wrote: > Erling Wegger Linde a écrit : >> >> On Mon, Jun 16, 2008 at 5:54 PM, Peter Keane <[EMAIL PROTECTED]> >> wrote: >>> >>> In our project, similarly to Connections, we use: >>> >>> <category term="public" scheme="http://.../category/visibility"/> >>> >>> I'd be v. interested to know what other folks do, esp. if there is an >>> opportunity for standardization. In fact, I could see real utility in >>> some semi-formal way to register category schemes and terms. As has been >>> noted [1][2], atom:category has the potential to be quite a powerful >>> exension mechanism, getting us a "description framework" without having >>> to use RDF (not that there's anything wrong w/ RDF ;-)). >> >>> From an outsider, just watching this list from time to time: Why not >> >> use RDF? Isn't RDF more general? Could these categories be used >> anywhere outside of Atom? >> >> I'm not very familiar with the atom:category(ies) so I ask: can >> atom:categories be referred to / used in a meaningful way with RDF? > > Unless I'm mistaken, they are URIs so yes I'd say a client that is built > with such purpose could make sense of those elements.
Ok, could you give me an example of such a URI? (I think I have never seen an atom:category scheme before..). What would happen if you use a URI to an RDF or N3 file instead? ( I think one should make it possible to use both ) > >> E.g. can you either use URIs e.g. "http://.../flags#private"; or >> "http://.../flags/private"; to refer to them? Or could one create a# Atom to >> turtle xslt 2.0 generated by the really excellent xq2xsl script from the >> above XQuery. It retains a small dependency on Saxon. Test it out by >> translating Tim Bray's Atom feed using the W3C XSLT 2.0 service. >> general approach for using GRDDL ( >> http://research.talis.com/2006/grddl-wg/primer ) ? If you cannot, then >> I think this approach will be a dead end in a few years, agree? > > No I don't. Could you explain? > > I actually believe atom:category elements have the basic minimum that > someone may be looking for when adding a context to a resource. Let's say that using RDF and other Semantic Web technologies is getting more widely used. It is my opinion that this will happen sooner or later. If the atom:category approach isn't very compatible with such Semantic Web technologies, then I think people will not want to use it in the future. I mean, what information are you trying to add with atom:category elements? Are you trying to state the scope/domain of an Atom feed? Where will these category schemes/URIs be stored? Does someone create public schemes for everyone to use? Or do everyone just create their own schemes? How can you tell if a scheme is related to another scheme? E.g. a concept in a scheme is a subclass of another concept in another scheme? Could you express something like that? I think that ontologies describing all sorts of concept will be created for many domains, independent of whether they are used with the Atom format/protocol or not. Why should one need to create such schemes for a specific format/protocol if they can be shared and reused other places using RDF? > >> >> If there is one thing I think Atom is missing is a standardized way to >> combine it with RDF. Why not try to use RDF when you can, not try to >> avoid it? > > I don't understand what you mean by standardizing RDF with Atom. If someone > is interested in using RDF inside Atom can't they just do so already? Or do > you mean mapping the Atom model to the RDF data model? > > Looking at the GRRDL Primer specification it looks like it's a > transformation mechanism of a source document to a RDF representation. Is > there any showstopper from the RDF community to already create such > specification for consuming Atom documents? What I mean is: it is probably easy to transform Atom elements such as <summary>, <author> etc. to RDF as they do not change! However, I wonder how the atom:category schemes looks like, and if it is possible to create a general GRRDL script that will transform them into something meaningful in RDF. But please give me an example of an atom:category scheme. > > More fundamentally my *personal* issue here is that RDF has chosen a > top-down approach by saying "We needed a mechanism to enable meaningful > consuming of heterogeneous resources" so they defined RDF and gradually > offered simpler mechanisms to allow the profane to join the RDF world (like > GRRDL Primer). Atom on the other hand has followed a bottom-up path and > been about "Let's have the lowest entry barrier we can and gradually add > more complex mechanisms if they are needed". But because the base format is > well defined you can make an Atom document already meaningful on its own. Yes, the simplicity and the RESTfullness of the Atom format/protocol is what I love about it. I just suggest that the atom:category approach is way out of scope! You have to draw the line somewhere I think, and I think atom:categories is far across that line, as it would be much easier to use RDF here. I think it is impossible to make the atom:category approach powerful enough (as users will probably demand more features in the future) without actually making it more similar to RDF, and that would be a waste of time and energy, as you could just have used RDF from the beginning. I think at least one should be able to use both. > > - Sylvain > > > > >> >> Thanks, >> - Erling >> >>> --peter keane >>> >>> [1]http://torrez.us/archives/2006/05/25/447/ >>> [2]http://www.majordojo.com/2006/05/overloading-atomcategory.php#c12851 >>> >>> >>> On Mon, Jun 16, 2008 at 08:01:38AM -0700, James M Snell wrote: >>>> >>>> In Connections, we've adopted a category based "flags" mechanisms, e.g. >>>> >>>> <category scheme="http://.../flags"; term="private" /> >>>> >>>> It is essentially a boolean flag. If the term "private" is included in >>>> the >>>> entry, the entry is private, otherwise it's not. The "flags" scheme >>>> contains a number of other types of flags relevant to the entry. So >>>> far, >>>> this has worked reasonably well for us. >>>> >>>> - James >>>> >>>> Bill de hOra wrote: >>>>> >>>>> Hi, >>>>> >>>>> are there any format or category extensions for stating view privacy or >>>>> moderation levels on content or feeds? ) >>>>> >>>>> Bill >>>>> >>>>> >>> >> >> >> > > -- Med vennlig hilsen Erling Wegger Linde
