|
Dear Pat,
Many thanks for sharing your hard won experience.
A few points:
1) ontologies that conform to (ii) and (iii) - I am probably very naive on this point - and many may argue about what is and what isn't an ontology - but I know of two current biomedical "ontologies" - one venerable and in use for about a decade - the other new and just coming out in beta - that meet these criteria as I read them. The former is The Gene Ontology. The latter is the Ontology of Biomedical Investigation (OBI). I'm only a user of the first, but am involved as a coordinating developer on the latter which is written in OWL and much more formally specified than GO CURRENTLY is. Both seek to conform to (ii) and (iii) and could certainly document the extent to which they do. As I'm sure you know, GO has a very large community both of developers, curators, and users, and has gradually brought the utility of ontologies to the biomedical research at-large, even though one can still come up with an argument for why GO is not an ontology. It certainly aspires to be, provides many of the advantages ontologies provide (e.g., classification), and is improving on its formal, well-founded nature as we speak (or type).
2) OBO and the OBO Foundry - by definition, OBO Foundry ontologies are expected to conform to (ii) and (iii). Many other biomedical ontologies in the OBO space do seek to conform and make very labor intensive, best faith efforts to do so
3) Biomedical ontologies - a special case? - There are two primary uses ontologies are used for in biomedical informatics: (a) focussed logic-driven, data-centric inferencing in a tightly constrained domain; (b) very broadly scoped annotation to a shared semantic framework. There are other apps (e.g., data federation and mediation, ontology-primed text mining and NLP, etc.), but these two applications dominate. I would say that for (a) above, (ii) and (iii) are neither necessary or sufficient. A great deal of that work is done with small, one-off, custom made ontologies. For the latter, there is little hope that task can either scale or the results be maintained without a shared foundation and without accepting the input of all comers via something like the loosely specified processes implied in (ii) and (iii). I'm primarily involved in (b), so that is why (ii) and (iii) are important to me.
4) Cathedral vs Bazaar - In the Linux/Open Source "Bazaar" model, the whole process would collapse if there were not a foundation shared by the majority of the community (the kernel). Even kernel extensions can cause all sorts of grief, not to mention the library dependency issues that have given rise to the thriving Linux distro players. That's just for Linux. Other Open Source OSes have different kernels, but each of these function as a shared foundation for the communities that develop to them - and use them.
Cheers, Bill On Mar 18, 2008, at 11:57 AM, Pat Hayes wrote: Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
boundary="----_=_NextPart_001_01C888D8.E73F7E40"
Dear Pat,
For clarification, I don't object to (ii) and (iii), but I do think that these should not be required. Insisting on any conditions on the process that gave rise to the ontology adds a considerable burden both to the cost of creating an ontology and to the task of checking its credentials, and is completely irrelevant to the users of the ontology. Very few, if any, extant published ontologies fully meet conditions (ii) and (iii) above. Ontologies are not standards: they are more analogous to pieces of software.
MW: I have to disagree with this. Ontologies are much more analogous to the specifications that software is made from. Standards often form part of that specification, and even user requirements are usually agreed amongst the users or their representatives. On the other hand I quite agree that few ontologies (as far as I know only those developed as data models in ISO like ISO 15926) have been developed this way. On the whole, I think it is a statement about the maturity of ontology as a discipline that people like yourself would consider that an open process for development was such a hard thing to achieve in practice. It implies that getting consensus at present is hard to impossible.
Its certainly a lot harder (and slower) than just writing the damn thing. Common Logic wouldn't exist now if it had been left to a committee. It got written because I convened a small hand-picked group and we agreed to get it done. My experience, which admittedly may be less extensive than yours, is that this is always how it gets done, in practice: some small subgroup of the committee just gets on with it and gets the thing written. As for 'open process' the openness of the ISO Common Logic process was a joke, whereas the W3C process in fact took place in public, and was strongly influenced by input from many sources (not just within the W3C), actively sought public comments, input from other WGs and so on. Exhausting, in fact.
As for maturity, while you may be right, I don't see any reason why being more mature is likely to produce more consensus on ontological issues. They seem to be permanent and intransigent. This is one reason I see ontologies as being very much more of a pick-your-favorite business than anything like standard-setting. A bazaar rather than a ministry.
Pat
Regards
Matthew West
Reference Data Architecture and Standards Manager
Shell International Petroleum Company Limited
Registered in England and Wales
Registered number: 621148
Registered office: Shell Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Mobile: +44 7796 336538
Email: matthew.west@xxxxxxxxx
http://www.shell.com
http://www.matthew-west.org.uk/
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2008/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2008
Community Portal: http://ontolog.cim3.net/
--
_________________________________________________________________ Ontological Engineer work: (610) 457-0443 Biomedical Informatics Research Network (BIRN) and National Center for Microscopy & Imaging Research (NCMIR) Dept. of Neuroscience, School of Medicine University of California, San Diego 9500 Gilman Drive La Jolla, CA 92093
Please note my email has recently changed
|
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2008/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2008
Community Portal: http://ontolog.cim3.net/ (01)
|