Wiki Classification Schemes [Was: Page rename crashed second time around]
Murray Altheim <[email protected]>
| Newsgroups | gmane.comp.java.jspwiki.user |
|---|---|
| Message-ID | <[email protected]> |
Claudia Frers wrote:
> Thx Dirk ..that was the idea but i was concerned about the long names needed
> to achieve this goal...but i guess that there is no "better" way.
Claudia,
It's common on wikis (absent any alternative) to attempt to overload
the wiki page name with classification information, especially given
that we typically have a flat namespace (i.e., all names exist within
the same context or scope; another way to put it is that the scope
is undefined).
There are two ways to deal with this:
1. come up with a wiki page naming scheme, publish it, and then
conform to it (and get all the wiki's users to conform as well)
2. stop using wiki page names for categorisation
Obviously the second is a more compelling approach, but the native
wiki solution of creating "Category*" links is similarly problematic,
as it has again a flat namespace.
There's a number of good books on library categorisation that I could
recommend, suffice it to say that there are many types of categorisation.
The most compelling of them is called Faceted Classification (FC), where
rather than enumerate a set of classes or categories, they are created
not by naming them but synthetically through the combination of what
are called "facets" (properties, characteristics, attributes, etc.).
So that rather than create an enumerated classification called e.g.,
"Extinct American Blue Butterflies" one would simply assign facets
to the entity such as "Extinct", "American", "Butterfly", "Blue", with
each of those facets (in 'classical" FC) being part of a hierarchy of
facets (e.g., "American" would be in a hierarchy of Place, "Blue" in
a hierarchy of physical properties, etc.). A search via facets would
produce a result set of all those entities having those assigned
facets. An advantage here is that if a given set of entities gets too
large it's always possible to create more facets to further differen-
tiate them, unlike an enumerated classification where fitting 5000
more entities between 299.514 and 299.515 might get problematic. With
FC it's not an issue. The facets are referred to by name, and are
drawn from a fixed set, called a "controlled vocabulary".
But even with a controlled vocabulary, the problem *again* is that
using names to identify facets we are still working within a flat
namespace. "American" is both a colloquial term for people from the
USA and the name of a hemisphere, "Blue" is both a colour and an
emotion, etc. There are ways around that to create namespaces or
enumerate facets, but not simple ways, and any enumeration harks
back to similar problems. The issue of what's called "subject
identity" is actually very subtle and complex, and commonly over-
simplified in most systems (including the "Semantic Web" world of
OWL and Description Logics).
Because most wiki users are not trained ontological engineers (who
are hardly immune to making profound mistakes and designing flawed
systems), we can either attempt to design and enforce a controlled
vocabulary as well as train our users in its proper usage (a
decidedly difficult or intractable task) or discard the very idea of
linguistic formality in favour of by-design informal system. That's
kinda what tags and folksonomies attempt to do, and there's been a
ton of work done in recent years, mostly by people with no background
whatsoever in epistemology, computational linguistics, knowledge
representation, formal ontological systems, or similarly daunting
fields. Lacking that knowledge we end up with very similar problems
cropping up.
So what to do?
I can't claim to have any solution for certain, but I've been working
on one for awhile now. It involves FC, "informal logical assertions",
and it's completely untested on a population of users. It's also not
available yet. But when it is, it will be compatible with (i.e.,
functionally based on) JSPWiki.
Suffice it to say this is hardly a simple issue, but it is one that
libraries have been dealing with effectively for a very long time.
Not sure if this was helpful, me just burning off some energy from a
very good sushi dinner... mmm
Murray
...........................................................................
Murray Altheim <murray07 at altheim.com> === = =
http://www.altheim.com/murray/ = = ===
SGML Grease Monkey, Banjo Player, Wantanabe Zen Monk = = = =
Boundless wind and moon - the eye within eyes,
Inexhaustible heaven and earth - the light beyond light,
The willow dark, the flower bright - ten thousand houses,
Knock at any door - there's one who will respond.
-- The Blue Cliff Record