Re: Improving classification of spams
Booker Bense <[email protected]> Mon, 13 Jan 2003 09:15:46 -0800 (PST)
| Newsgroups | gmane.mail.ifile.general |
|---|---|
| Message-ID | <[email protected]> |
On Mon, 13 Jan 2003, Clemens Fischer wrote: > Booker Bense <[email protected]>: > > > - Also, I throw away the .idata file and reindex things every > > couple of weeks. Not sure if this has any effect or not. > > i think this doesn't do much good. how dod you get the feeling that > doing this is useful? > - Well, I mostly do it because I recently started a new job and keep changing the way I organize my email. However, the more I think about it, the more useful I think it is. Email lists tend to focus on specific topics and then drop them after a while, they are very "bursty" data. What you really want to capture is the statistics of the current content of the email list, not it's past. - It might be very interesting to track this kind of data through time and build some underlying model of what you expect an email list "spectra[1]" to look like. - Booker C. Bense [1]- I'm sure that's the wrong word to describe what I mean, but I've been hanging around a lot of physicists lately.