[Sigia-l] That stuff we do, ontologically speaking

[Alexander Johannesen]

Hi Ruth,

On Feb 10, 2008 7:25 AM, Ruth Kaufman <ruth.kaufman at gmail.com> wrote:
> Have you considered making this ontology about content types rather than
> page types?

Sure, I've done those for years and those are easy. :) The problem
isn't so much the content classification as the classification of the
work we do as IAs. Customers aren't all that interested in content
classification when what they want / expect are a sitemap, and
certainly not at the early stages. This ontological endevour is just
to map out what we usually put on those prototypes and sitemaps (and,
for those who really are interested, it's going to be used as a
meta-ontology you place on other ontologies [for example content
typing] to map out and auto-magically create full prototypes / working
websites of them).

> The notion of page is disintegrating on the web.

Hmm, not sure I fully agree with that. I think I understand what
you're saying, but the notion of a page will always exist as long as a
URL resource is called that, so even in a widget / content type driven
world, there's still pages that display them. Yes, there's more focus
on typing these days, but the venerable "page" won't die because of
it. I guess what you're saying is that the old notion of content type
isn't strongly linked to whatever page you're seeing, which I can
agree with.

> The intent was to use this taxonomy as a set of allowed values for the HTML
> metatag DC.

Well, if you want to speak about stuff that's disintegrating on the
web today, I'd choose DC. :)

> Type across all of ibm.com's published web pages, including
> application-created pages (4 million or so in all). The primary categories
> were:
>
>    - Site orientation (includes entry pages, home pages, landing pages,
>    search pages/results, error pages, help content, etc.)
>    - Account information (includes registration pages, user profiles,
>    order history, etc.)
>    - Purchase transaction enablement (for e-commerce stuff)
>    - Technology information (white papers, etc.)
>    - Business information (thought leadership, etc.)
>    - Offering information (product descriptions, etc.)
>    - Business partner enablement information (sales and marketing
>    materials for business partners)
>    - User assistance information (guides and such for users of IBM's
>    products)
>    - Support (technical support docs for troubleshooting, etc.)
>    - Reference (for indexes, footnotes, glossaries, etc. -- literally
>    reference material)
>    - Communities (for blogs, wikis, etc.)
>    - Events (events, obviously)
>    - IBM information (all the corporate stuff and news about IBM)
>    - Business partner information (partner directories, etc. -- for
>    customers)
>    - Conversion pages (for things like "thank you" pages)

Excellent list, which, incidentally, has both of the types of
ontological expressions we're talking about. First you've got the type
of the page type, then the page type examples in brackets. You've got
a different breakdown of the types of page types, which is much better
than mine which is solely based on old ideas. Thanks, that's very
helpful, and I'll revise and update in a day or so.

> The entire taxonomy was up to 4 levels deep (i.e., each branch could be no
> more than 4 levels deep), so that's the 2 top levels. The interesting part
> of this was getting people (many stake holders) to understand and accept
> that this was literally a taxonomy, not a navigation scheme or object
> model.

Which is why I hold a lot of courses and workshops on typeification
and ontology work before I work closely with the people in any
organization. The shift to classification thinking is more difficult
than what you would think given our language as complex classification
models. :)

> Another design point you may find helpful was that we tried, as much as
> possible, to separate content type (our meaning was closest to "genre") from
> presentation format and MIME type. In some cases this was impossible, such
> as with the notion of "webcast". A lot is baked into that word. We actually
> had webcast twice, but with different unique id's -- once as a media-based
> event and once under event replay.

Good advice, and although I'm not doing this for a specific client nor
a specific technology, I have to add that MIME types in certain good
architectures *can* match your content ontology (for example when
designing truly RESTful systems), but yes, normally MIME content type
is used as media type.

> Finally, we tried to craft this taxonomy in the spirit of RDF -- meaning, in
> a way that in the future, we could break it down into triples.

In theory, triples have little to do with it, and I'd advice against
using triplets as a way to model anything (as you can decompose
complex models into triplets later if you must). You can express
pretty much the same stuff in RDF and Topic Maps (there's a few
notable differences in identity notions) or any model of choice.

> You're calling your work an ontology. I suppose you're planning to establish
> relationships among entities beyond the simple hierarchy? (You mentioned
> topic maps.) Sounds interesting.

One step at the time, but yes, extending the ontology to cover
validation, cardinality, semantic expressions and so forth would be
cool. But first I'll put something together that can bridge the work
IAs do and the world of ontologies.

I should add that what I'm intending to do here and what I'll end up
using it for first isn't really dependent on each other. The ontology
is generic and can be used by all who wants to (I'll find a cute place
to host it). I'll first use this as part of a tutorial for the Topic
Maps 2008 conference in April where you can take any model (Topic Map,
RDF, etc), make a small merging map between the model and this
ontology, and out the other end comes a full website where navigation
is tied to content models.

Anyways, thank you for your input; it was very valuable.


Regards,

Alex
-- 
---------------------------------------------------------------------------
 Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------