- From: Alexander Johannesen <alexander.johannesen@gmail.com>
- Date: Tue, 21 Feb 2012 15:02:49 +1100
- To: Tim Berners-Lee <timbl@w3.org>
- Cc: Rob Styles <rs@kasabi.com>, Semantic Web <semantic-web@w3.org>
Hi Tim, Thanks for your reply, and I appreciate the sentiment and the pointers. I've been doing SemWeb stuff for almost 15 years, and have written a couple of inference engines on top of resolvers to create complex mixed models, so all of that is fine, and I understand and accept that no other model is going to work on the scale of the web (I'm a die-hard RESTafarian). But epistemology is my game, and IMHO the authorship of a single ontology has a tendency to give a more precise shorter answer over large sets of data than a host of several small, especially considering the issue of ontology impedance which still isn't anywhere near being solved in a generic way, even though W3C tries very hard to consolidate. (I tried for many years to involve the library sector in this debate) I recently had to clean up someone else's mess because people treated dc:title as foaf:title. :) > If this sounds like a mess, it is in fact the pattern which > actually is optimal Hmm, I wouldn't call it optimal, but certainly my best option. When you accept REST, you are making compromises to your ideal solution for What Works [TM], and I'm all down with that. > I have written this up in http://www.w3.org/DesignIssues/Fractal.html > and specifically http://www.w3.org/DesignIssues/Fractal.html#TCO I can't shake the feeling that there's a big assumption here that other people are creating models that a) are correct and b) agrees with you and c) semantically fits my data (or vice versa)? I understand what you're saying, and I think in an ideal world you're probably right (and it's a great principle to follow, and I've often taught this point, especially to enterprise environments), I just don't think I can agree that this is actually how it happens (depending on the need for precision). Shared and mixed ontologies will always be tied to the quality of modelling within them - even the tiniest little thing! - and the semantic impedance between them, rather than it being done at all. (And the difficulty of getting these things wrong might in itself provide an answer to why I can't seem to find many Intranet ontologies out there) I can also make a case that a mixed model might be the very thing that guarantees you a future failure given certain contexts, but I take your point. > I hope that helps you hit a happy balance between build and buy, > and adjust it with time as necessary. I'll probably mix and edit a few good ontologies, and define a fork of them as its own ontology, although the Gist ontology looks like a good upper / mid-ontology for it which I can extend. I just thought there'd be more in that space. :) Kind regards, Alex -- Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps --- http://shelter.nu/blog/ ---------------------------------------------- ------------------ http://www.google.com/profiles/alexander.johannesen ---
Received on Tuesday, 21 February 2012 04:03:17 UTC