Am 14.03.2011 11:13, schrieb Richard Cyganiak:
On 14 Mar 2011, at 09:15, Bob Ferris wrote:
2. The selection of ontologies listed is, to say the best, often biased or
partly a random choice. I do not know any repository that
- lists more non-toy ontologies than abandoned PhD project prototypes.
I don't want to take a concrete position here, however, every ontology
development has somewhere its starting point and is there usually not so
popular. Nevertheless, the ontology design can be a good one, too. For that
reason, why should be abandon these approach and brand them as evil?
The point in re-using a vocabulary or ontology is this: one joins a community
of data publishers and re-users who have agreed on certain shared terms for
shared concepts.
The abandoned PhD project type of ontology or vocabulary has no community
around it. Therefore, one gains very little by re-using it.
This is why it's so important to involve multiple stakeholders from the start,
and get feedback from real data owners and data users along the development
process. That's the first and perhaps most important step in the process that
you called “ontology marketing” elsewhere in this thread.
Yes, you are absolutely right. However, not every ontology designer has
the power or reputation to get valuable stakeholders on board (I think,
I made my personal experience in that area* ;) ). So, I can only repeat
myself: PhD-project-born ontologies have not to be bad per se, or?
Banning them a priori is a rather prejudiced approach in my mind. When I
have to choose an ontology, I try to initially review all available**
ontologies independent whether they have their origin in a PhD project
or design by a big industry consortium.
Bad design decisions can be made everywhere - in the small-grouped PhD
project or that one with a huge industry community behind. I think every
ontology has the chance to get somehow famous, or?
The ontology with huge stakeholder community in the background is damned
to get popular and the little-sized-project-born ontology has the
freedom to get accepted somewhere and somehow.
Regarding "ontology marking", I especially try to address the following
issues:
- the ontology shall be discoverable, even by fuzzy requests (that is
why, the tagging approach that is followed by Schemapedia is a quite
good one) and by general purpose search engines alá Google
- the ontology specification shall be provided in as much as possible
and appropriated serialization formats, e.g., RDF/N3, XHTML+RDFa,
RDF/JSON, RDF/XML
- the ontology shall be published with a good (interlinked)
documentation, incl. illustrating examples, graphics of its structure,
related ontologies, etc. (ideally everything at least available in
XHTML+RDFa)
- the ontology shall be evolvable by a community, incl. issue trackers,
mailing lists, etc.
Cheers,
Bob
*) No feedback is also a kind of feedback
**) every ontology I can find that might be somehow appropriated to
fulfil my addressed purpose somehow