Using Emergent Classification as a Starting (Not End) Point. http://www.adaptivepath.com/publications/essays/
From elearningpost comes mention of this useful article by Peter Merholz (some may remember him from 'peterme' days, one of my earliest regular blog reads).
D'Arcy,
King and I had been trading emails a few weeks back on the value of
emergent classifications systems like those seen in Flickr for use in
learning object repositories. Clearly, the idea is getting a bit of
play, at least within the blogosphere.
What troubled me was that some of the current executions seemed a
little bit like a baby/bathwater thing - yes, emergent classification
systems are interesting and reflect actual users' language usage, but
they are also problematic - in being flattened, they do not have the
depth (and the corresponding teaching ability) that hierarchical
taxononmies can offer their users, and are also plagued with some of
the problems Merholz points to. I mean, have you ever actually tried to
find something you know should be there but didn't know the
classification for, (as opposed to just serendipidously browsing), in
an flattened keyword system?
Instead, I think Merholz describes better than I did in my emails to
D'Arcy and King what I think we should be looking towards - using
'emergent' temrs as the basis for creating connections between terms
users actually use, as the basis for continual refinement of more
complicated, less flattened, taxonomies.
How would this actually work - at the very least I think it could
show up in things like 'type ahead' functionality that tries to
complete the term you are entering based on previous 'emergenet' terms,
or else asking the user to confirm whether they were using a term in
one sense or another after they have submitted their choice. - SWL[EdTechPost]
-- I think you are missing the power of search cimbined with usage
information - the google way to usefulness - is a workable organic
model that maps human behavior with some extra facilitation with
classifications/hierarchies of links. The place where cogent
hierarchies will work is in personal search bots that to a degree map
the hierarchial category structure of the human for whom the search is
being performed. A server system can only go so far with "common
language" that is necessarily built on past usage of a shared language
such as English. -- suggestion: keep the xml-rpc gateway open for
assisted searching because the user's machine is the proximal client.
-- BL
10:33:34 AM  Google It!.
|
