- From: Sean B. Palmer <sean@miscoranda.com>
- Date: Wed, 29 Nov 2006 12:07:33 +0000
- To: www-tag@w3.org
The standardizedFieldValues-51 crisp summary: > Since short strings are scarce resources shared by the global > community, fair and open processes should be used to manage > them. A pattern that I'd like to see more of is > > 1. start with a URI for a new term, > 2. if it picks up steam, introduce a synonym that is a short string > thru a fair/open process. > > Lately I'm seeing quite the opposite. I recently came across this problem again from a different angle. My use case is a Semantic Web UI, which is trying to display property values in as compact a form as possible to the user. If you have a screenful of this: <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://xmlns.com/foaf/0.1/Person> <http://xmlns.com/foaf/0.1/homepage> <http://inamidst.com/sbp/> It's hard to see what's going on. In XML and Turtle/Notation3, you can shorten property values using QNames... rdf:type foaf:Person; foaf:homepage <http://inamidst.com/sbp/> It's a lot easier to see what the properties are, especially if they were to then be colour coded too ("foaf" seems yellow to me, and "rdfs" red...), but here come the problems: * The QNames aren't a part of the model, they're part of the syntax. The information is lost after parsing, so unless you have a specialised parser looking at xmlns and @prefix, you lose it. * Even if you do collect the data, you're relying on conventions. This is the standardizedFieldValues-51 issue in a nutshell. The interface becomes confusing as soon as another language called "FOAF" is invented independently somewhere. My first suggestion is that the process for resolving standardFieldValues-51 be decentralised. I've already spoken about using reverse domain names as a human readable but universal identifer space for RDF profile hinting: http://lists.w3.org/Archives/Public/semantic-web/2006Nov/0134 RDF Profiles in Link Values; by Sean B. Palmer Date: Fri, 17 Nov 2006 13:20:42 +0000 The semi-solution comes from thinking around ways of squeezing them into URI space somehow (an exercise left to the reader). Consider: http://xmlns.com/foaf/0.1/homepage com.xmlns.foaf.homepage net.rdf.foaf.homepage ont.foaf.homepage foaf:homepage Or secondly, from my point of view as a Semantic Web UI gendankenexperimental designer, I'd be happy with some kind of algorithm that lets one compress an HTTP URI in some well-understood way to make it more human readable yet still globally unique. On the other hand I can't think of a particularly good one. Thirdly, we could just come up with yet another global identifier space. What have we learned from DNS? That charging for names commoditises them, that charging for the mere rental of names leads to people worrying about persistence, and that people don't manage their own spaces very well. We have domain name speculators cleaning up by registering all possible names for the next pope, and things like that. We have people losing their domains to Network Solutions. But establishing such a space as truly global is such a bother that we're probably stuck with what we've got. If you could, however, redesign DNS from scratch, how would you do it? For five billion people to have even one human readable identifier each in their own language... there are 354 million people with English as a first language according to Wikipedia, but I'm not sure how big the space of orthographically/phonologically valid combinations would have to be to accomodate that amount of people; and moreover this doesn't solve the social problem of trademarks and the such; who deserves the really decent names? First come, first served is kinda harsh to the people just being born now; and adding new top level domains isn't really proving chic enough. If the Semantic Web is to have a well managed process for coming up with really short identifiers from extremely well used URIs like foaf:homepage, it might be up to an implementation to power it, just as Google gives great credence to people who come first in the results for various terms. Yours handwavingly, -- Sean B. Palmer, http://inamidst.com/sbp/
Received on Wednesday, 29 November 2006 12:07:51 UTC