- From: Danny Ayers <danny.ayers@gmail.com>
- Date: Fri, 18 Aug 2006 10:06:27 +0200
- To: public-grddl-wg <public-grddl-wg@w3.org>
A problem with applying GRDDL to existing microformat data is that the majority of documents don't have profile URIs. Try [1]. Clearly this is in part this is due to few of the microformats having been assigned profile URIs, presumably also through publishers not using the URIs that have been assigned. This does have immediate, practical impact on the reach of GRDDL - aside from the generic XHTML data-view, the transformations are associated with these URIs, so the lack of profile URI means with the current mechanisms, none of the domain-specific data is automatically available in RDF. I believe this points to two issues that should probably concern this WG: 1. how to get URIs associated with the microformats; 2. how to deal with documents which lack profile URIs. 1. Given that the microformats community has agreed in principle [2] that each microformat should have a URI, this becomes the question of how to encourage a follow-through on this, getting the URIs minted. Ryan King offered [3] the following reasons for why there aren't more profile URIs: [[ 1. We want some degree of stability first. 2. There are more important things to work on and the microformats can grow and evolve without profiles. (if you disagree, you're more than welcome to volunteer to help with the profile URIs :D) ]] Neither of these seems a blocker for getting URIs minted. At [2] there are some candidate domains - microformats.org, www.w3.org, xmdp.org to which I'd add purl.org I honestly don't know the best approach, but as microformat authoring tools are appearing fairly rapidly, any action that might encourage the microformats folks to get URIs in place soon would be a good idea. 2. Dealing with microformat data which lacks a profile URI is a different story. Right now it involves essentially scraping for attribute strings like "vevent". Seems to me there are two general strategies: raise the status of these strings to registered identifiers (presumably e.g. class="vevent" is relatively unlikely to appear in non-microformat docs); find a way to express the information that any data extracted from the document was done so without the license the profile URIs provides. Arguably in practice it might make sense to simply apply the appropriate transformations an gather the triples anyhow. DanC noted (in chat) the use or republication of such triples would be with the understanding that the authority for such triples would then be on the user or republisher. To me this seems undesirable as it involves information loss (the existence or otherwise of the original profile license). But given that many microformat docs will need pre-processing to XHTML, and capturing a chain of such transformations as a GRDDL profile seems overkill, I'm not sure this is something worth losing sleep over... Cheers, Danny. [1] http://kitchen.technorati.com/search/ [2] http://microformats.org/wiki/profile-uris [3] http://microformats.org/discuss/mail/microformats-discuss/2006-July/004769.html -- http://dannyayers.com
Received on Friday, 18 August 2006 08:06:46 UTC