with much help from Vadim, i got going yesterday evening well enough
to do some training, and then this morning i turned up filtering.
it's working well, but in its current form it needs a bit of work before
it can be unleashed on the unsuspecting world.
0) a setup wizard might be a good idea; it could encourage people to
use dspam in the mythical "right" way (e.g., don't send spam to
the trash, but to a quarantine folder for review before disposal,
that sort of thing.)
1) if configure doesn't see sqlite, even if --with-dspam is specified,
it builds without giving any indication that dspam was skipped, and
the menu entries in Message|Spam still show up and when used, don't
generate error messages.
2) it would be nice if there were a way to corpus train from
Message|Spam; the existing entries are for error training which
isn't quite the same thing.
3) could we have error training on the right-click menu for
messages?
4) it would be good to add a button for clearing statistics on
the Edit|Spam Filters menu for dspam, as after you're satisfied
with the training, you usually are well advised to clear stats,
as errors in training tend to skew totals for a Very Long Time
after dspam moves into production mode.
5) it'd be nice if the displayed stats computed percentages
for false positives, false negatives, etc.
6) there also should be a mechanism in Edit|Spam Filters
for purging the database. perhaps there should be some
automation, e.g. mahogany periodically checks for the
last purge date and if it's been a while, it runs one. this
could be an option, e.g. AutoPurge. also, purge after
training completion is important, as the early databases
get quite large and once in production, they can be
shrunk a lot.
7) corpus training from a mbx file with 1000+ messages
takes a while (at least, on my 750mhz linux laptop.) some
sort of progress bar would be good, but this will likely
require an API enhancement in libdspam. i can take this
to jonz.
8) at one point, jonz was talking about setting up a dspam
fed DNSBL. i will check with him on its status, it'd be nice
if we could use mahogany dspam implementations to
contribute to such a project. jon's conception avoids
the traps of spamcop (the spamcop bl being a good idea
with some serious execution issues that render it unusable
for many of us.)
richard
--
Richard Welty [EMAIL PROTECTED]
Averill Park Networking 518-573-7592
Java, PHP, PostgreSQL, Unix, Linux, IP Network Engineering, Security
-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 -
digital self defense, top technical experts, no vendor pitches,
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
Mahogany-Developers mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mahogany-developers
