[topicmapmail] PSI standardization

[Murray Altheim]

Shyam Peri wrote:
 >> Are Gulbrandsen wrote:
> Dear Sir,
>
> The need for this is definitely there, but there has to be a certain 
> level of standardization for a PSI to be usable.
> 
>> I think individual organizations would like to host their own local PSI 
>> sets for their local ontologies, to be in control. At the same time they 
>> would like to map parts of their ontologies to standards based PSIs to 
>> be able to easily exchange and merge topic maps, and because it's a bit 
>> stupid to make your own definition of every data modelling and ontology 
>> concept.

I agree strongly with this idea, but I don't see that many domain-specific
PSI sets are going to need to be mapped by their domain-specific users into
larger ontologies -- this is probably going to be done by others who have
the interest and expertise. And such mappings are very prone to errors, as
once out of a specific domain, the language used is very different from
other domains, i.e., there is no universal set of definitions for terms,
and often terms are in violent disagreement between communities (I don't
need to cite examples like Palestine or issues like abortion in much detail
to make this point). You can't standardize a concept like "life" if you
can't obtain agreement across communities, but you can standardize it within
a community. It's often shared agreements on language and concepts that
define a community.

> But there seems to be no such governing body who takes care of these standardization.
> Most the PSI which are being identified and modeled, are being confined to the specific 
> application developers.
> I appreciate the classfications which you have modeled on the PSI.

There is no governing body on the planet suitable for standardization
of all PSI sets.

There are a number of concepts that have been conflated repeatedly in this
discussion:

## Registry           a listing of things (online or otherwise)

Now, I can't think of any individual website or organization that would
*not* be suitable as a PSI registry. Anybody can create a list. The value
of that list will be in its currency, accuracy, and breadth and depth of
coverage, or in the case of specific domains, how well it covers that
domain. There are no general-purpose registries anywhere in the world;
all registries of things are domain- or topic-specific.

## Repository         a source of documents (or things)

Likewise, a repository can be set up by anyone willing to devote the
time, energy, and disk space. There may be legal issues in adding an
item to the repository without consent, depending on the license of
the item. The value of a repository is its stability, management, and
how well-accepted it is within its domain. There are no general-purpose
repositories anywhere in the world; all repositories of things are
domain- or topic-specific. The closest thing to a general-purpose
repository would be a museum, and even here we find there are specific
types of museums.

## Standardization    the act of standardizing something, which requires
                       a standards body such as ISO, IEEE, IETF, ANSI, etc.

There are standards bodies big and small, devoted to standardizing things
within their domain. Some are more general than others, but even the big
ones (such as ISO or ANSI) have a very specific scope or requirements
prior to consideration of a standard. Most have formal working groups and
processes.

## Recommendation     the act of some body recommending something, such
                       as might be called "best practices". E.g., this is
                       the most that the W3C can do (hence, the final stage
                       of a draft is a "W3C Recommendation") since it is
                       an industry consortium, not a standards body

Industry consortia are put together by companies with money and their
reputation relies in large part upon the "network effect", i.e., how
many players are involved, or which specific players are involved. In
some industries, a consortium has been set up that one or more of the
big players refuses to take part, limiting the value of the consortium.
E.g., if Microsoft backed out of the W3C, its value would plummet.
Processes and working groups exist in consortia, sometimes well-managed,
sometimes poorly-managed.

Now, we can go around about "standardization" of PSIs and other things,
but really, there's little point doing so here. For every PSI set devoted
to a specific domain, say, medicine, it's likely that the PSI set will be
developed and maintained within the medical community, and standardized
by some medical community standards body (or consortium, as appropriate).

What I think we're trying to do here (and this is all my opinion, as
should be obvious), is develop methodologies and best practices for the
creation and publication of PSI sets. We aren't in the business of trying
to standardize those sets. Now, some of the things that are directly
applicable to Topic Maps themselves we can perhaps standardize through
the existing ISO process, but I believe most of what is being discussed
here are PSI sets for specific, non-TM domains, and it would be highly
inappropriate for the SC34 WG8 working group to begin standardizing PSI
sets for say, the Spanish Knitting Association. Let Spanish knitters
work it out. We just need to give them technical guidance on how to do
it themselves.

Murray

...........................................................................
Murray Altheim                         http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK                    .

     Why is it that they impeached Clinton when he lied about having sex
     with a consenting adult, but they aren't talking about impeaching
     Bush over having led the world into war by lying about the presence
     of WMDs in Iraq? Is the US more squeamish about sex than war?