[Sigia-l] The category of "Miscellaneous"

[Jonathan Broad]

I hate to interject this late in the thread, but there are alternatives 
brewing out there that break down the "pure algo" v. "manual structure" 
dichotomy that's been richocheting around.

Google took the world by storm for good reason.  Their algorithm, while 
magnificently elegant, is hardly fool-proof or universally optimal.  A 
good way of looking at it is that It just taps into an accidentally 
excellent source of metadata about a web page--what other web pages 
that page esteems with a link.  It scales really well, but has some 
serious weaknesses with regards to domain-specific search.

"Best bets" work in a lot of cases.  But as user interests shift and 
datasets grow and change, they're difficult to maintain.  Simple 
directories, for their part, are of limited utility because of the 
likelihood of dead-ends and re-tracing, so we should shed no tears for 
the passing of their heyday.

Endeca has an excellent commercial implementation of faceted 
search/browsing.  The Flamenco project headed by Marti Hearst of 
Berkeley's Info School has an open-sourced implementation of a similar 
interface.

For one thing, that kind of interface eliminates the dead-ends of 
directories.  For another, it re-homes relevant structure dynamically 
in response to a user query.  It's not web-scale, because without 
*some* structure it's really just a neat and fast clustering algorithm. 
  But information attracts structure rather naturally.

So I guess my question is, what's wrong with a dynamically generated 
interface to structured information?  Structuring information is good, 
but not at the expense of dynamic and flexible access.  But it need not 
be an either/or.   A compromise is in order.

Keyword/faceted-navigation hybrids are, for the enterprise, the best 
way forward I think.  It reminds me of what my father once described as 
his educational ideal: military montessori.  As he put it: "A little 
goose-stepping, and a lot of free play".

Jonathan
jonathan at relativepath.org

On Jun 16, 2004, at 12:04 AM, Listera wrote:

> Christina Wodtke:
>
>> Humans are not terribly articulate creatures.
>
> The most articulate in the universe that we know of.
>
>> Browse is a query formation tool.
>
> So is the iterative process of search-and-refine.
>
>> And sometimes a human providing a representative set of results works 
>> better
>> than allowing popularity to define the meaning of the term.
>
> That's a simplistic definition of, say, Google.
>
> Historical note: people by the millions voted with their clicks to 
> move from
> directory browsing (Yahoo) to algorithmic search (Google), not the 
> other way
> around. Something more cost-effective than the latter might come along 
> one
> day, but I haven't seen one yet.
>
> Ziya
> Nullius in Verba
>
>
>
> ------------
> When replying, please *trim your post* as much as possible.
> *Plain text, please; NO Attachments
>
> Searchable list archive:   http://www.info-arch.org/lists/sigia-l/
> ________________________________________
> Sigia-l mailing list -- post to: Sigia-l at asis.org
> Changes to subscription: http://mail.asis.org/mailman/listinfo/sigia-l
>
>

-----------------------------------
Jonathan Broad
jonathan at relativepath.org