[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

To RDF or not to RDF, that is the (perennial) question


On Thursday, January 6, 2005, at 07:50 AM, Danny Ayers wrote: 
The potential benefits of RSS 1.0-like extensibility have often been
stated on-list. One regular counter-argument has been along the lines
of "I don't see what that buys us" (usually from Dare ;-), which has
usually prompted yet another rewording of what it does buy us.

Part of the problem is that the benefits of RDF are often explained in language that, while I assume it's clear to people conversant with RDF, is only vaguely comprehensible, and that with effort, to those of us who aren't. From time to time I catch a glimpse of the value of RDF, but I doubt that I've ever really seen the big picture. I see two pages on the wiki addressing the RDF or not question (http://www.intertwingly.net/wiki/pie/NoToRdf and http://www.intertwingly.net/wiki/pie/XmlAndRdf), but neither strikes me as being a good "the big picture of RDF benefits for non-RDF speakers" page.

Another part of the problem is that some people simply don't care about the benefits that RDF offers. While I'm sure the benefits are more than worth cost of RDF syntax to those who do care about those benefits, they are not for those who don't. Is there a way to reconcile the needs of the two groups?

I see four possible futures:

1) Atom does not use RDF and is not defined and constrained in a way that makes it easy to convert to RDF (including elements from other namespaces). This would make Atom significantly less useful to those who care about RDF's benefits.

2) Atom does not use RDF, but it is defined and constrained in ways that make it easy to convert to RDF (including extensions). This should be at least minimally acceptable to pretty much everyone.

3) Atom uses RDF, adding a lot of syntactic overhead and/or variability to its structure. This would make Atom significantly less appealing to those who aren't conversant with RDF and/or using RDF-aware tools.

4) Atom uses RDF, but the way documents are structured is constrained to a subset of how RDF can be structured so that it doesn't require RDF-aware tools to process, and the amount of overhead added by the RDF syntax is minimal. This should be at least minimally acceptable to pretty much everyone.

Which of #1 and #2 are we closest to right now? Is #2 or #4 achievable? If so, which is preferable? I'd like to aim for #2, but if the overhead and obfuscation in the eyes of non-RDF speakers of #4 is truly minimal (I suspect that RDF speakers and non-RDF speakers would have very different definitions of "truly minimal"), I'd have no complaints about going that route. Also, in either case, if working toward a particular solution significantly increases our time to completion, that's a major strike against it.