On 01/04/12 18:01, M. Fioretti wrote:
On Sun, April 1, 2012 4:41 pm, Mike Scott wrote:
Actually, while I remember a disagreement, I'm not sure myself of the
details. I think it might have been about using some sort of engine to
guess a likely answer to unsubscribed queries - I vaguely seem to recall
you didn't approve:
Yes, now I remember, thanks. I stepped in because you suggested a bayesian
autoresponder, and I explained why in my opinion it wouldn't work in this
case. But it was a purely technical disagreement, nothing personal. Heck,
Nothing personal taken. We're all allowed - I hope - our opinions.
I did actually start categorising list responses at one time, just to
see if a Bayesian system might work (it was only a particular example).
But it was taking too much time, so I dropped the project and chucked
the data. Ah well.
....
Bayesian filters cannot work in this case because they assume that the
Maybe. I don't know, and thought it might be worth trying. There's
usually /some/ key word or phrase that gives the game away.
sender knows what he's talking about. And it's not needed at all, if the
sender receives answers from real humans anyway. But of course the copy
One thing I was trying to reduce was the endless repetition of the same
answers. People have better things to do, I hope, than tell endless
numbers of others the same thing over and over -- 'you unsubscribe
by.......'. Boring to do, boring to see on list.
--
Mike Scott
Harlow, Essex, England
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
