Adrian and Paola, (01)
That is a critical issue that we must face: (02)
AW> On the one hand, a useful ontology is supposed to be "right",
> in the sense that it will not be necessary or desirable to change
> it much. On the other hand, an ontology is supposed to capture
> some aspects of the changeable real world. But, if an ontology
> is used to support, say, interoperation of 100 legacy systems
> in an SOA architecture, then making a change could break many
> of the interoperations. (03)
Any proposed ontology, no matter how good, will never be adopted
unless it can accommodate legacy systems and support a smooth
transition from where we are to where we want to go. (04)
AW> So, a key aspect of using ontologies in practical situation
> would appear to be to have some reliable change management method
> or better, change management software. The software would have to
> be able to run automated regression tests over the entire collection
> of SOA of legacy systems after each change to the ontology. (05)
Something along those lines is essential. The task of a standards
body is not to implement the software, but to define an agreed
methodology with firm interfaces. Ideally, the definition should
be a refinement and extension of an already implemented prototype
that has proved to be useful for a wide range of applications. (06)
AW> Also, the software would need a user interface that business
> folks could understand, without mediation by IT specialists. (07)
That would be highly desirable. But that would not be something
that should be in the standard, because good human factors are
extremely difficult to do well, and they can be revolutionized
overnight by a brilliant innovation -- just consider what the
Wii, the iPod, and the iPhone did to those industries. (08)
For the standard, I believe that we should build on existing
standards, such as the Metadata Registry, and on well defined
mathematical systems. Robert Kent's IFF system, for example,
has been suggested, and I believe that it would be an excellent
basis. However, the full details of category theory, etc., are
more than even IT specialists should have to learn. (09)
That is why I believe some agency should sponsor efforts to build
prototypes that include the high-powered math under the covers.
And those prototypes (preferably in a design competition) should
be built as open-source *research* projects. Then any commercial
vendor could adopt and adapt the winning system in an industrial-
strength version. (010)
For example, relational databases use very sophisticated indexing
mechanisms, but the IT specialist can ask a question in SQL and
a non-specialist could use an English-like interface without
being aware of the underlying math. (011)
PDM> reality is made of quantifiable and not quantifiable subsets
> therefore
>
> a better (less fictional) model of reality would include the
> product of calculus, as well as some representation of what cannot
> be quantified and a possible model of the interaction between the two (012)
What I would recommend is that the ontology include a library of
mathematical theories, any of which could be used to support any
ontology. For things that aren't quantifiable, such as feelings
and taste, very simple math might be used, such as saying "62% of
the respondents said they prefer chocolate ice cream to strawberry." (013)
For subjects that can support detailed calculations, different
mathematical theories might be used for different applications.
Ordinary Euclidean geometry, for example, is sufficiently accurate
for any measurements on earth. But the gravity of the sun warps
the space nearby, and the theory of general relativity is necessary
to compute the orbit of Mercury. (014)
The library of mathematical theories would therefore include all
the versions of geometry (and many other branches of math), and
any particular version could be used as appropriate. (015)
John (016)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (017)
|