[topicmapmail] what about a "non-hierarchical chart" ?

[Bernard VATANT]

Lance, Jack

The "no hierarchy" slogan was deliberately over-provocative. You had to
read "no *absolute* hierachy ", and it does not mean I would gladly hang
the last taxonomist with the guts of the last librarian :o)
Sure you need trees in zoology, laws and everywhere. What I fear is the
paradigm of the "Universal Knowledge Tree", dmoz or SUO-style (reminds me
of some ancient story about such a Tree).
The n-connected tree approach you suggest is used somehow in
multi-dimensional analysis. You can always formally speaking identify a
non-hierarchical chart in many ways as a tree, starting from any node.
That's the point : in *many* ways. And I agree the real tricky  work is
making these many coexist in some coherent way.

Anyway, it does not answer my zen enigma about where I can put my "no
hierarchy" group in real-world (?) egroups single hierachical tree (with
cross links OK, but basically and semantically a tree).
When I was dmoz editor, I never did index my own domain name for the same
reason : it would not fit in any subcategory, dealing with astronomy,
Hautes-Alpes, web indexation ... well, sort of weird multidimensional
object (as its webmaster himself). That's when I began to have some doubts
about dmoz concept.

----- Message d'origine -----
De : Lance Otis <lanceo@mail.apptechsys.com>
À : 'Jack Park' <jackpark@verticalnet.com>; 'Bernard VATANT'
<b.vatant@wanadoo.fr>; <topicmapmail@infoloom.com>
Cc : 'Howard Liu' <hliu@verticalnet.com>
Envoyé : vendredi 20 octobre 2000 17:50
Objet : RE: [topicmapmail] what about a "non-hierarchical chart" ?


> Here are my thoughts on your comments:
>
> Consider a model of a domain that is a two dimensional hierarchy with
depth
> and width:  is it not possible that an object not yet linked within the
> hierarchy (for example: 'no hierarchy'), could be linked in a third
> dimension to many items in the hierarchy? Further, is it not possible
that
> two 2-dimensional hierarchies could intersect along a third dimension?
Could
> a hierarchy be considered to be no more than a selection of useful nodes,
> and links between these nodes, from an n-dimensional domain of node
> relationships?
>
> The n-dimensional data domain being modeled may consist of many nodes
> associated with other nodes via links, commonly called arcs.  The art of
> modeling (building a set of consistent hierarchies) lies in including
those
> associations and nodes in the model that are of interest, while omitting
> irrelevant nodes and associations.  This requires an architecture that
can
> accommodate sets of associated objects as trees.  These trees are
> associations selected from the network of possible associations in the
> domain being modeled.  Multiple node association trees may be necessary
to
> form the basis of the model representing the domain.
>
> For each association tree selected from the domain being modeled,
inverted
> association trees from any point in the hierarchy, should be generated so
> that any object can be directly accessed as the central node of its own
> hierarchy. Thus, instead of a non-hierarchical chart, we would have many
> interconnected hierarchies.
>
>
> -----Original Message-----
> From: topicmapmail-admin@infoloom.com
> [mailto:topicmapmail-admin@infoloom.com]On Behalf Of Jack Park
> Sent: Wednesday, October 18, 2000 10:17 AM
> To: Bernard VATANT; topicmapmail@infoloom.com
> Cc: Howard Liu
> Subject: Re: [topicmapmail] what about a "non-hierarchical chart" ?
>
>
> Make it so...I'm in :o)
> I am wondering if there might be a bit of *non-Aristotelian* thinking
here.
> If we move away from hierarchical structure, taxonomic categorization,
and
> so forth, what will we get?