Re: RDFa and Microformats

On 12 Sep 2008, at 10:46, Toby A Inkster wrote:

> RDFa is "just" a representation of RDF. And microformats can  
> already be parsed as RDF - that's the point of GRDDL. Despite the  
> fact that XSLT is a horrible, horrible abomination, I think that  
> GRDDL, not RDFa, is probably the best hope for bringing  
> microformats into the "upper case Semantic Web".

To clarify my point, I'm not saying that I think that Microformats 
+GRDDL are the best way forward for the semantic web. I think RDFa is  
great, and would recommend it as the first option to consider if you  
want to add semantics to web pages. What I mean is that for those  
existing pages that are already using Microformats, and indeed future  
pages that will be written with Microformats, I think GRDDL  
represents the best way to pull that data into the upper case  
Semantic Web - bloating RDFa with algorithms to parse all existing  
Microformats would be a mistake. Indeed, the guidelines on how to  
parse existing Microformats right now are far from comprehensive -  
some areas are extremely murky and grey.

> this is a new model (inheritance design pattern) used in only  
> haudio at the moment so its experimental I'm not sure it works though?


It works - I've implemented it. It is a great boon for hAudio - if  
you're describing an album, you don't want to have to re-list the  
contributors for every single track - you want them to be able to be  
inherited from the album. Though hAudio gives you the flexibility to  
be able to list track-specific contributors too, which is good.

But what's good and useful for hAudio is less useful in the general  
case for other formats. With microformat parsing rules like this  
being defined specifically for each microformat, and indeed each  
property, microformats are able to hone themselves to be perfectly  
designed and beautifully compact formats for their own particular  
topic. But it also means that writing a universal parsing algorithm  
for all microformats is virtually impossible. Even if you manage to  
cover all the special cases in current formats, new formats may  
introduce other interesting (and useful in their particular case!)  
nuances. So you have a choice between three possibilities:

1. You choose to write your algorithm for current microformats and  
resign yourself to the fact that it won't cope with future microformats;
2. You write your algorithm for current microformats and update it  
every few months; or
3. You write your algorithm for current microformats, and then  
constrain the development of future formats so that they must work  
with the existing algorithm.

Given that RDFa is trying to be fairly universal and allow virtually  
any data to be represented, #1 is out of the option. #2 is also out  
of the question, as RDFa is supposed to be a fairly stable  
specification. #3 would be a great shame as you'd miss out on one of  
the best thing about microformats - the fact that each has tonnes of  
useful shortcuts for publishers.

If RDFa and Microformats can be brought closer together, then that's  
great, but we shouldn't compromise RDFa's simple, predictable  
processing model, nor the great flexibility of microformarts.

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>

Received on Friday, 12 September 2008 23:01:48 UTC