Re: Improving the list (was: Can we please dekookify the list?)

Timothy Bates (tbates@karri.bhs.mq.edu.au)
Tue, 02 Mar 1999 15:31:35 +1100

> THIS MESSAGE IS IN MIME FORMAT. Since your mail reader does not understand this format, some or all of this message may not be legible.

--MS_Mac_OE_3003233495_1416529_MIME_Part Content-type: text/plain; charset="US-ASCII" Content-transfer-encoding: 7bit

Anders Sandberg suggested

> I have been thinking of running some program through the list archives
> to do data mining, for example build sociograms (who responds to
> whom?), find threads where someone participates, who glorifies who and
> perhaps extract clusters of discussion. Maybe the way of improving
> quality of the list lies in making archives more powerful?

Certainly. The archive server could then add simply meta-tag information to mails based on some kind of discriminant function analysis, or a weighting on 5-6 neural net vectors.

clients could then run private scripts like this simple example (in Applescript 'cause I am a Mac head)

property myWeightings:{0.5,0.1,-0.7,0.9,0.01,}

set thisMailsWeights to (extract meta weights(mail_headers)) as list

set myRanking to 0
set n to 1
repeat with myWeight in myWeightings

set myRanking to myWeight * item n of thisMailsWeights set n to n + 1
end

if myRanking is less than 3 then

mark the current mail as read
set the color of the current mail to "light gray" end if

cheers,
tim

--MS_Mac_OE_3003233495_1416529_MIME_Part Content-type: text/html; charset="US-ASCII" Content-transfer-encoding: quoted-printable Re: Improving the list (was: Can we please dekookify the list?)</TIT= LE> </HEAD> <BODY BGCOLOR=3D"#FFFFFF"> Anders Sandberg suggested<BR> > I have been thinking of running some program through the list archives= <BR> > to do data mining, for example build sociograms (who responds to<BR> > whom?), find threads where someone participates, who glorifies who and= <BR> > perhaps extract clusters of discussion. Maybe the way of improving<BR> > quality of the list lies in making archives more powerful?<BR> <BR> Certainly. The archive server could then add simply meta-tag information to= mails based on some kind of discriminant function analysis, or a weighting = on 5-6 neural net vectors.<BR> <BR> clients could then run private scripts like this simple example (in Applesc= ript 'cause I am a Mac head)<BR> <BR> <B>property</B> myWeightings:{0.5,0.1,-0.7,0.9,0.01,}<BR> <BR> <B>set</B> thisMailsWeights to (extract meta weights(mail_headers)) as list= <BR> <BR> <B>set</B> myRanking to 0<BR> <B>set</B> n to 1<BR> <B>repeat</B> with myWeight <B>in</B> myWeightings<BR>    <B>set</B> myRanking to myWeight * item n <B>of</B> thisM= ailsWeights<BR>    <B>set</B> n to n + 1<BR> <B>end</B> <BR> <BR> <B>if</B> myRanking is <B>less than</B> 3 <B>then</B> <BR>    mark the current mail as read<BR>    <B>set</B> the color of the current mail to "light g= ray"<BR> <B>end if<BR> </B><BR> cheers,<BR> tim<BR> </BODY> </HTML> <p> --MS_Mac_OE_3003233495_1416529_MIME_Part-- <!-- body="end" --> <p> <ul> <!-- next="start" --> <li><a href="2662.html">[ Next ]</a><a href="2660.html">[ Previous ]</a> <b>Maybe in reply to:</b> <a href="2449.html">Alexander 'Sasha' Chislenko</a> <!-- nextthread="start" --> </ul> </body></html>