On Oct 14, 2011, at 4:42 AM, "Lin Clark" <lin.w.clark@gmail.com<mailto:lin.w.clark@gmail.com>> wrote: That's part of the idea that Richard brought up after having thought about it, and he was the one that suggested that there would be approximately three patterns. I assume that one would be hash URIs. I don't know what other patterns he has identified, if there are any. I can think of three: http://a/b#type + name => http://a/b#name http://a/b/type + name => http://a/b/name http://a/b/type + name => http://a/b/type#name The third is something like originally recommended in Hixie's RDF conversion. It also seems at odds with the vast majority of vocabulary uses. This can be supported, however, either through the use of full URIs in @itemprop, or by using a name such as "type#name". I'd really rather not depend on a registry for doing vocabulary specific conversion. Gregg -Lin On Fri, Oct 14, 2011 at 12:33 PM, Jeni Tennison <jeni@jenitennison.com<mailto:jeni@jenitennison.com>> wrote: Lin, On 14 Oct 2011, at 11:28, Lin Clark wrote: > On Fri, Oct 14, 2011 at 9:52 AM, Philip Jägenstedt <philipj@opera.com<mailto:philipj@opera.com>> wrote: > This registry could also define a small number of patterns for forming the URI from the combination of the itemtype and the token. For example, one pattern would be to take the domain, without the path to the itemtype, and append the itemprop token as schema.org<http://schema.org> does. There would probably be around 3 patterns. Then, a vocabulary could be registered as following one of those patterns. What were the other patterns, out of interest? Jeni -- Jeni Tennison http://www.jenitennison.com -- Lin Clark DERI, NUI Galway<http://www.deri.ie/> lin-clark.com<http://lin-clark.com> twitter.com/linclark<http://twitter.com/linclark>
Received on Saturday, 15 October 2011 14:50:03 UTC