Search and feature extraction collections defined; common shot reference coming soon

[Paul Over <over@nist.gov>]

Hi,

Two items...

1) Collections: -------------------------------------------------

We've posted a final list of files randomly divided so that
about 40 hrs are designated for the search collection and
just under 30 hrs for the feature extraction collection. Two
additional files from the Internet Archive were added last
week as they became available. The other 15 we were looking 
for but have not been able to get as of today, have been dropped 
from the list. 

This is THE list. Please make sure you are training and testing 
using the correct files. There are a lot, some have come and gone
(maybe without your noticing), and it's way too easy to invalidate
your results by missing a file or having an extra one from an
older version. So do your checking now. :-)

<http://www-nlpir.nist.gov/projects/t2002v/collection.txt>

We've heard from only one group about assistance in getting
a copy of the files. We assume everyone else has downloaded them
from the Web or will do so VERY soon.


2) Common shot reference:---------------------------------------

The group at Dublin City University has kindly volunteered to
provide a common shot reference for the entire identified
collection. They believe they can do this by May 20th.
Although we had thought about soliciting such references
from multiple groups and somehow blending them for fairness's
sake, we think getting the reference done quickly is more
important. Many things depend on it. So we plan to use just 
the DCU reference as THE common shot reference for the search 
and feature extraction tasks.

- Paul & Ramazan


-- 
Paul Over - Retrieval Group
	    Information Access Division
	    Information Technology Laboratory
	    National Institute of Standards and Technology
	    Bldg. 225  Rm. A211  (Mailstop 8940)
	    Gaithersburg, MD  20899-8940   USA
	    Voice: 301 975-6784    Fax: 301 975-5287