- From: Xiaoshu Wang <wangxiao@musc.edu>
- Date: Tue, 18 Nov 2008 23:32:47 +0000
- To: John Graybeal <graybeal@mbari.org>
- CC: Alan Ruttenberg <alanruttenberg@gmail.com>, Michael F Uschold <uschold@gmail.com>, "semantic-web@w3.org" <semantic-web@w3.org>, "aldo.gangemi@gmail.com" <aldo.gangemi@gmail.com>, Conor Shankey <cshankey@reinvent.com>, Peter Mika <pmika@yahoo-inc.com>, Ora Lassila <ora.lassila@nokia.com>, "Pan, Dr Jeff Z." <jeff.z.pan@abdn.ac.uk>, Tim Berners-Lee <timbl@csail.mit.edu>, Frank van Harmelen <Frank.van.Harmelen@cs.vu.nl>, "sean.bechhofer@manchester.ac.uk" <sean.bechhofer@manchester.ac.uk>, "obo-format@lists.sourceforge.net" <obo-format@lists.sourceforge.net>
The issue is not about *version policy* but about ontology's *design and deployment*. We should take lessons from software engineering, where we are all trying to remove the code dependency by carefully refactoring code to handle change. Ontology engineering is not much different because our knowledge about the world is bound to change. But we would like our URI to be stable. Hence, the concern of ontology engineering is to manage the logic-dependence through careful ontology modularization. I have discussed this issue and offered some solutions in a paper that I presented at DILS 08[1]. 1. Ontology Design Principles and Normalization Techniques in the Web. http://www.inesc-id.pt/ficheiros/publicacoes/4799.pdf Xiaoshu John Graybeal wrote: > On Nov 9, 2008, at 8:50 AM, Alan Ruttenberg wrote: > > >> I should point out that within the Open Biomedical Ontologies there is >> an explicit policy of *not* changing URIs as new versions of the >> ontology are released - for one thing that would be impractical - some >> of them are updated daily. Rather there is a policy on deprecation - >> terms that are deprecated are marked as such and kept in the ontology >> so as not to leave dangling pointers. >> > > Term deprecation works for me. And I think there does need to be a > single (i.e., unversioned) URI for each term, that always reflects the > 'latest semantics' of a given term. My view is that we're really > creating a dictionary that maps a string to a definition (crudely > put)*, and yes the semantics/definition for a term may change over > time, so that should be reflected in the 'latest' expression of the > term (the one that the 'latest semantics' URI represents). > > But I think there also needs to be an appreciation that the exact > concept associated with that term is evolving over time, and each time > it explicitly evolves, that is a slightly (or hugely, sometimes) > different conceptualization. It isn't at all impractical to create a > new URI for each such change, even if it changes every minute, at > least that I can see. > > What I like about this scenario is that a user can use either concept > for the term -- the very specific versioned one, or the unversioned > 'latest' one -- according to what they want to express. I don't see > why both concepts shouldn't be available. > > I tend toward the opinion (thanks to Michael U's arguments) that > changing the versioned URI should only occur when something about the > term is explicitly changed -- new versions of the vocabulary, in and > of themselves, are not sufficient. > > John > > > * So Peter's defense of visible names resonates for me. > > > -------------- > John Graybeal <mailto:graybeal@mbari.org> -- 831-775-1956 > Monterey Bay Aquarium Research Institute > Marine Metadata Interoperability Project: http://marinemetadata.org > > >
Received on Tuesday, 18 November 2008 23:33:41 UTC