Corpora: A new project: Open Mind Word Expert

Rada Mihalcea rada at utdallas.edu
Mon May 6 07:04:09 UTC 2002


We would like to inform you about a recently started project, "Open
Mind Word Expert", which is part of the Open Mind initiative, and aims at
creating large sense tagged corpora with the help of Internet users.

The OMWE system is running at http://teach-computers.org.

Since the official launch of the system at the beginning of April, we
collected about 6000 tagged instances at virtually no cost. Based on
previous experiences with Open Mind type of approaches, we expect a
significant increase in time of the amount of labeled data collected
through this site.

We would like to point out two important features that many of you might
find of interest:

1. The site is designed such that it can be used as a teaching aid. We are
already aware of several courses that had assignments or exams based on
this site, where students get a good grasp on the various meanings that
a word might have by tagging several word examples. The 'project' feature
allows the project administrator to monitor the tagging done by the users
registered with that particular project.

2. The data collected through this site is made freely available for
research purposes. You can download and use the corpus for your own
experiments in Word Sense Disambiguation, Machine Learning, or any other
research related experiments.

We feel that this project can create a valuable resource for the research
community.  A project of this nature, however, succeeds because of its
participants, not its designers.  If you feel that this is a worthwhile
project, please spread the word.  It would be particularly helpful if
you link to us from a web site you maintain.

So far, we have been profiled at:
http://coolsiteoftheday.com (May 6, 2002)
http://topwebgames.com/cat.asp?name=Board
http://www.gamecase.nl/strategy/index.html
http://gamespotter.com/showcategory.php?Cat=9&OrderBy=Title&Begin=177

"Open Mind Word Expert"
Tim Chklovski, MIT
Rada Mihalcea, UTD



More information about the Corpora mailing list