W3C home > Mailing lists > Public > semantic-web@w3.org > January 2006

Re: How about?

From: Joshua Tauberer <tauberer@for.net>
Date: Tue, 10 Jan 2006 17:00:06 -0500
Message-ID: <43C42E66.6030302@for.net>
To: love26@gorge.net
CC: SWIG <semantic-web@w3.org>

William Loughborough wrote:
> I can approach thousands of people and ask "what's the name of the tune 
> that goes 'dah-dah-dah-BOOM?'" and they will instantly respond 
> "Beethoven's Fifth".
> 
> When will our machine indexing, etc. stuff allow me to do that with a 
> google-type search and a microphone?

The association of 'dah-dah-dah-BOOM' and B's 5th is an interesting example.

People probably give the correct answer to this particular question not
necessarily because the tune actually sounds like that and they know it
(e.g. I can make the association without you even providing a sound
file -- the text 'dah-dah-dah-BOOM', which has little in common with the
original, was enough), but more because there's a cultural thing that
we all know associating something pretty abstract (three low-volume
things and then a high-volume/low-pitch thing) with B's 5th.  I'm sure
there are lots of people that can make the association (maybe not by
name) without having heard the symphony.

So then the question is, which would you want the machine indexing to 
model -- the abstract cultural thing, or some sort of tone matching?

(Just an amusing aside-- On CNN just now, as the reporter was talking 
about wire taps, an XML file scrolling around was being used as a 
graphic.  I guess if there has to be wire taps, it might as well be 
encoded in XML.  Good for exchange between intelligence agencies, right?)

-- 
- Joshua Tauberer

http://taubz.for.net

** Nothing Unreal Exists **
Received on Tuesday, 10 January 2006 22:00:34 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:47:11 UTC