RE: OEP] Constrained hierarchies and meta-classes

Aldo

I'm happy to see you are almost as confused as I can be by documentalists' world :))

> 1) documentalists maintain heading hierarchies (I assume they are a
> sort of "topic")

Yes, "sort of". They call it a "taxonomy" (sigh) or, in French "Plan de classement".
In a project I'm currently involved, we deal with such a taxonomy for "People"
publications (challenging stuff, huge legacy, alas) in a major press group.

We have a heading (or category) hierarchy like :

Person
	. Public Life
		.. Awards
	. Private Life
		.. Family
		.. Holidays
		.. Couple
		.. Pets

For each heading, there is attached information such as
	- description of types (classes?) of relevant documents
	- types (classes?) of related knowledge base objects
	- naming convention (for the indexed folders)
	- attached keywords (in a relevant Thesaurus)

So, for the taxonomy manager, headings are (sort of) (individual?) topics with attached
properties, organized in a hierarchy.

> 2) documentalists manipulate folders (I assume folders are locations
> of books or documents)

Yes. Used to be (and still sometimes are) paper folders, and now electronic ones. A folder
can be the container of actual document files, or of some electronic proxies, e.g. URIs.

> 3) documentalists declare some relation between folders ("subFolderOf")

Yes. You do that any way you like in your hard disk folders. But in our use case
documentalists have to follow the constraints defined by the taxonomy. Not only the
hierarchy, but also the naming conventions etc ... and of course put the right documents
in the right folders.

So you will get something like the following hierarchy of folders :

Britney S
	. Britney S - Public Life
		.. Britney S Awards

etc ...

> 4) ???categories are defined wrt to the subFolder relation???

Tricky things begin here. You can say "Britney S - Public Life" is an instance of a
generic class "Folder". But it's also specifically related to the generic "Public Life"
heading. What is the nature of that relation, that is the question. Even if the taxonomy
manager does not think about headings as classes, the most natural (?) way to express this
relation is to consider "Public Life" as a class of which "Britney S - Public Life" is an
instance ...

> 5) ???a category is a subClassOf Folder???

... seems the logical conclusion of the above ... (category = heading).
If "Public Life" and the like are subclasses of Folder, you can describe "a posteriori"
the heading hierarchy by restrictions on the subFolder relation, as clearly explained in
your previous message. But this will nevertheless sound strange to the taxonomy manager
...

> I am lost. Maybe you could provide an excerpt of a folder set, and an
> excerpt of a heading hierarchy, so that our intuition gets closer to
> the point.

Hope the above makes for that. I was lost for a while, too :))

> As usual, I do not care much about what representation primitives
> should be used, until I do understand the intended meaning of the
> experts. OK, documentalist use informal structures, but before
> deciding on what formal structure we can suggest, we need to know
> what they want to talk about, and for what purpose.

Well, it is in fact a recursive process. Providing accurate language at the right moment
generally help people to figure out better what they are about. All the (pedagogic) point
is to bring this language at the right point. Is not it what we are about?

<snip/>

> I interpret "representing at best the legacy" as a good encoding of
> the intended meaning/use, not as a (possibly awkward) encoding of the
> original structure.

Agreed.

> If we are able to understand the semantics of the legacy, we can even
> reproduce the original format or syntax, but being aware of the
> underlying semantics.

... or propose an alternative one (simpler, more generic, more interoperable, less
ambiguous ...)

<snip/>

> Give me some example to judge on awkwardness ...

Imagine the above example with 10 000 + headings to manage. Does it scale?

<snip/>

> A mereology as an ontology of part is any theory that includes part
> relations in its signature, and provides axioms for that relation wrt
> a specified domain. There is a lot of curious or fancy stuff out
> there, but let's try to stay focused ...

Indeed :))

Ciao

Bernard

Bernard Vatant
Senior Consultant
Knowledge Engineering
Mondeca - www.mondeca.com
bernard.vatant@mondeca.com

Received on Thursday, 10 June 2004 09:05:49 UTC