- From: Danny Ayers <danny.ayers@gmail.com>
- Date: Wed, 9 Aug 2006 23:32:08 +0200
- To: public-grddl-wg@w3.org
Typically I remembered a few things after the meeting (please change subject on any reply to specific point) - * atom/owl use case? * live bookmarks use case? I reckon it would be good to cover these spaces for a variety of reasons, though I can't think of any suitable stories offhand. I'll ask around (#swig I guess). * validation/testing I can't see very far ahead as to what would be possible, but I'm sure there's something useful. Might be worth collating the various blog posts regarding microformat validation. One plan I got partway towards experimenting with (it's waiting on me revising the xslt) was to set up an atom2rdf.xsl, loading the result alongside the atom/owl ontology and doing DL consistency checking. If that proved not-much-use, plan B was to try rule-based validation along the lines of Schematron2 (rules infer statements regarding validity from the doc, sparql over them pulls out error messages) . One further item mentioned tangentially with the hReview use case is what to do where documents are believed to be GRDDL but lack profile URIs. This came up when I was trying to write a demo around microformats, few of the instance docs in the wild have the URIs, though the intention is clear from the use of the HTML class names (and use of Technorati's pinger). A possible approach, which DanC didn't seem keen on (in #swig the other day, and I couldn't really come up with a concrete proposal) was to use a property to flag that a graph was non-authoritative, i.e. not explicitly licensed by the publisher to express the triples. For want of a better scenario, consider a search engine, where some of the results would be derived from truly intentional microformat docs and others effectively scraped for well-known attributes. The end user of the search engine might well want to know whether the original publisher had applied the profile stamp. Cheers, Danny. -- http://dannyayers.com
Received on Wednesday, 9 August 2006 21:32:17 UTC