"Gordon V. Cormack"  writes:


>> The NIST on-line system for submitting public
>> corpus results will be available mid-August
>> (recall that the deadline for submission is
>> August 23).


Actually, things are ready now.

You can submit your public corpus results at
http://ir.nist.gov/trecsubmit/spam-results.html

There is a simple script available at
http://trec.nist.gov/act_part/tools.html called
"check_spam_results.pl", but since this file comes right out of the
jig, it's mostly a sanity check.  It does check the number of fields
in the file (online task has four, active has five) and it does check
that you only score each file in the corpus once.

(I just posted it, so it might take a little while to appear on the
external web site.)

One thing that you need to be careful about is that it's easy to
submit the wrong result file.  The output of the jig does not include
the run identifier, so there's no way for us to check that you
submitted the right file (beyond being able to tell online from active
learning outputs).  The confirmation email that you receive from the
submission system includes an MD5 checksum of the file which you can
and should check against your own copy.

Ian



-- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.1.394 / Virus Database: 268.10.5/406 - Release Date: 02/08/2006