Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: converted to 1.6 markup

Rescore Mass-checks for Set 0 and Set 1

The set 0 and 1 mass-check runs for 3.0.0 are now finished! This page is obsolete

Here's the procedure you'll need to follow, if you wish to submit data for the rescoring run next time around:

First, send mail to <>, and ask for a rescore-submission account if you haven't already got one.

It's helpful, but not required, to have some or all of the helper applications installed:

  • the Mail::SPF::Query module
  • the Net::DNS module
  • Razor
  • DCC
  • Pyzor

If you're running nightly mass-checks, please feel free to disable them when running the rescore mass-check runs. Also, please note that the nightly submission accounts will work for rescore submissions as well.

Then run these commands:

No Format
  tar xvfz Mail-SpamAssassin-3.0.0-pre2.tar.gz
  cd Mail-SpamAssassin-3.0.0
  perl Makefile.PL < /dev/null; make

  cd masses
  mkdir spamassassin
  rm spamassassin/bayes*
  echo "use_bayes 0" > spamassassin/user_prefs
  echo "use_auto_whitelist 0" >> spamassassin/user_prefs
  rm ham.log spam.log

  ./mass-check --net -j 4 --restart=400 --after=1041397200 --all <targets>

<targets> is the list of directories, mboxes, etc., like
spam:dir:~/Mail/spam. See the comments at the top of "mass-check" for details.

This takes *ages* to run. -j 4 controls the number of processes to use; 4 should be OK for a single-processor machine, since most of the time they'll be waiting for network results to arrive. If you have adequate RAM and don't mind the load, you can use -j 6 or -j 8. There's not much benefit in going higher than -j 8.

The --after=1041397200 option tells mass-check to ignore messages older than 18 months ago (in this case January 1 2003). This is useful if your corpus has older messages intermingled with your newer messages.

If you have an unusual network layout, you may need to specify
trusted_networks and/or internal_networks in the spamassassin/user_prefs file. But SA should be able to infer it in most cases. If you get less than a 10% or 15% spam hit rate for RCVD_IN_XBL, then you might need to use these configuration parameters.

Once it finishes:

No Format
  USER="[whatever your username is]"
  RSYNC_PASSWORD="[whatever your password is]"

  rsync -CPcvuzb ham.log $$USER.log
  rsync -CPcvuzb spam.log $$USER.log

That's it! Then we do the bayes+nonet and bayes+net runs later on.

The results for this run will need to be in by Monday July 19th. If you're still running then, submit what you have so far and beg for more time. (wink)