Re: Archiving and access & analysis = (bo po mo fo?)


Subject: Re: Archiving and access & analysis = (bo po mo fo?)
From: Richard Wentk (richard@skydancer.com)
Date: Sun Oct 16 2005 - 16:00:02 EDT


At 19:24 16/10/2005, you wrote:
>At 10:36 -0400 2005/10/16, Ned Bouhalassa wrote:
>>On 15-Oct-05, at 11:56 PM, Kevin Austin wrote:
>>
>>>Such a project would face the same issues that all archiving has to,
>>>namely, how to sort and catalog so that things can be found.
>>
>>Keywords might be sufficient.
>
>
>In my experience, this is not "trivial" (*)

It's trivial in the sense that the perfect is the enemy of the good enough.

God-like powers of sonic deduction and analysis aren't required - just an
ability to search a probably quite small database of items looking for tags
that have a good chance of being common enough to be useful.

This is more or less how Google handles a much bigger database of images,
and it really is very good at finding things.

As for not finding soundwalks on Google - remarkably enough, I found a
Sydney soundwalk by typing in "Sydney" and "soundwalk". And an audio
example of a iib V I cadence by typing in "iib V I".

Considering that this collection will be *much* smaller, with a much less
noisy and ambiguous search space than the tens of billions of items indexed
by Google, I'd guess the chances of getting something useful out of keyword
searches would be good enough to make anything more complicated unnecessary.

Besides, once you have the data you can always reindex it later with a more
sophisticated automated scheme if it becomes necessary.

But if the idea never gets off the ground in the first place, you'll have
nothing to index at all.

Richard



This archive was generated by hypermail 2b27 : Sat Dec 22 2007 - 01:46:13 EST