General Text Matcher (GTM)

written by:
Ryan Green
Joseph P. Turian
I. Dan Melamed
Luke Shen
Ali Argyle
Ben Wellington
Daniel Galron

The latest version of this software is version 1.4

To be notified of upgrades, please sign up for the GTM-announce mailing list via the web-form at http://www.cs.nyu.edu/mailman/listinfo/gtm-announce. It's moderated and very low-volume.

Introduction

GTM measures the similarity between texts. To read more about using this software to automatically evaluate machine translation, please read Proteus technical report #03-005, Evaluation of Machine Translation and its Evaluation, available here. To learn more about our work, please visit the Proteus Project homepage.

GTM is written in Java, and is open source, released under the BSD license (available in the LICENSE file).

The Proteus Project is supported by grants and contracts from the National Science Foundation (NSF) and the Defense Advanced Research Projects Agency (DARPA), as well as by gifts from Sun Microsystems.

Demo

A simple applet demonstrating the core functionality of GTM is available at http://nlp.cs.nyu.edu/call_gtm.html.

Documentation

All documentation is in the README file, which is also included in the distribution.

Downloads

1.4 (January 2008)

fixed bug that occurs when processing blank lines
added an option to use a file to indicate lines to compare
added options for dumping hits and/or runs

1.3

Changed text option to interpret newlines as
segment separators, and number the segments
consecutively. Previously, all lines in a file
were glommed into one segment.
Made the SGML reader case insensitive

The integrity of the following files can be verified by checking the message digests. The files are signed with Ali's key.

Source TGZ: gtm-1.3-source.tgz (33 KB)

MD5 (gtm-1.3-source.tgz) = c040548cff1655560d2994b1e67aa4df
SHA1 (gtm-1.3-source.tgz) = 20815ce3e95e1b2dc531eaecde92325a71564b45
Signature: gtm-1.3-source.tgz.asc (Signed with Ali's key)

Source ZIP: gtm-1.3-source.zip (65 KB)

MD5 (gtm-1.3-source.zip) = e207d0837767d873f9b865f2d7389706
SHA1 (gtm-1.3-source.zip) = 0f20303f2623e42d450af54b0453b52f3fec2932
Signature: gtm-1.3-source.zip.asc (Signed with Ali's key)

Binary TGZ: gtm-1.3-binary.tgz (34 KB)

MD5 (gtm-1.3-binary.tgz) = 8a9f41c7f8e0d373425dd226d01ff25d
SHA1 (gtm-1.3-binary.tgz) = 62ece243241da4e0fdb34a1857eb64d662dad089
Signature: gtm-1.3-binary.tgz.asc (Signed with Ali's key)

Binary ZIP: gtm-1.3-binary.zip (64 KB)

MD5 (gtm-1.3-binary.zip) = 4cd6ceb967442ba1bb4dd1da5be6143e
SHA1 (gtm-1.3-binary.zip) = 863d9943da9ca17b534e3c651ffd19107148b551
Signature: gtm-1.3-binary.zip.asc (Signed with Ali's key)

1.2:

Fixed mispackaged jar in the binary package

1.1:

Plain text input can be chosen using -t.
Encoding is set to ISO8859-15
Non-standard ASCII can now be used.
Added Documentation about ISO8859-15.

Note: All files version 1.0 and older were signed with Joseph's key.

Source: gtm-1.1-source.tar.bz2 (26 KB)

MD5 (gtm-1.1-source.tar.bz2) = cfd5e5f564de3e0372e919f011325278
SHA1 (gtm-1.1-source.tar.bz2) = 5831af0018ebeb7c155272a63b46f726ba3324a1
Signature: gtm-1.1-source.tar.bz2.asc (Signed with Ali's key)

Binary: gtm-1.1-binary.zip (61 KB)

MD5 (gtm-1.1-binary.zip) = 9cd008dbbcc3359b69da6906d9664868
SHA1 (gtm-1.1-binary.zip) = bac71f8bccb25947eab62a7f4e59b8f2db2b3487
Signature: gtm-1.1-binary.zip.asc (Signed with Ali's key)

1.0:

Default exponent is now 1.0.
Tie-breaking is now deterministic.
Re-enabled the sanity check suite.
Removed legacy GPL library.
Pointer to TR 03-005.
Samples no longer include non-standard ASCII text.
Added documentation note about non-standard ASCII.

Source: gtm-1.0-source.tar.bz2 (26 KB)

MD5 (gtm-1.0-source.tar.bz2) = eb1cd7432ca0fdbc257497d029f3aa7b
SHA1 (gtm-1.0-source.tar.bz2) = c0b45aea8d0beb8070ab3d1ae9848d1d202b6aa6
Signature: gtm-1.0-source.tar.bz2.asc

Binary: gtm-1.0-binary.zip (61 KB)

MD5 (gtm-1.0-binary.zip) = 307d7cb8f3b271279311f2e999206313
SHA1 (gtm-1.0-binary.zip) = 5e3c9a3efab2f4e0b157bfe028c15fcda2dc46ff
Signature: gtm-1.0-binary.zip.asc

0.3:

Documentation updates.
Temporarily disabled the sanity check suite.

Source: gtm-0.3-source.tar.bz2 (2.0 MB)

MD5 (gtm-0.3-source.tar.bz2) = c72b014699e2a79cc52160bfe77321ba
SHA1 (gtm-0.3-source.tar.bz2) = 1015106ad94e5ef81a8d012c12c975d8def4db74
Signature: gtm-0.3-source.tar.bz2.asc

Binary: gtm-0.3-binary.zip (2.4 MB)

MD5 (gtm-0.3-binary.zip) = f9314a2715c2ff96a187855ceb47347d
SHA1 (gtm-0.3-binary.zip) = 975b8ecf7da0a956e6fce2bdb1a114c48fbbf45f
Signature: gtm-0.3-binary.zip.asc

0.2:

Verbose output can be enabled using "-v".
Sanity check suite included.

Source: gtm-0.2-source.tar.bz2 (2.0 MB)

MD5 (gtm-0.2-source.tar.bz2) = 8426f5f1c4e5b8ef04c6eeaed1e68fcc
SHA1 (gtm-0.2-source.tar.bz2) = 32cb1f3b966eeecb48416dea061e7392ddecbbec
Signature: gtm-0.2-source.tar.bz2.asc

Binary: gtm-0.2-binary.zip (2.4 MB)

MD5 (gtm-0.2-binary.zip) = f3add53619692b7d54608323d9159003
SHA1 (gtm-0.2-binary.zip) = 40df630a44035c606d4c6ef8d6e5b6bfa4bbc025
Signature: gtm-0.2-binary.zip.asc

0.1:

Initial release.
Scoring of segment and document matches.

Source: gtm-0.1-source.tar.bz2 (2.0 MB)

MD5 (gtm-0.1-source.tar.bz2) = e2643e71dad28b26668e2d9abd06a45b
SHA1 (gtm-0.1-source.tar.bz2) = 14011c6af853bc16a478c695e29bb4cd106745c3
Signature: gtm-0.1-source.tar.bz2.asc

Binary: gtm-0.1-binary.zip (2.4 MB)

MD5 (gtm-0.1-binary.zip) = 44dd83baaf40c230644a483669bda0f4
SHA1 (gtm-0.1-binary.zip) = 8f2b19c033c0bd8655329ab1b9bdffa749d9165c
Signature: gtm-0.1-binary.zip.asc

Feedback

Questions? Comments? Suggestions?
Bugs? Code patches? Feature requests?

This is beta software and we welcome all feedback.
Please email feedback to Joseph Turian and I. Dan Melamed.
Our email addresses are [lastname] 'at' cs 'dot' nyu 'dot' edu

index.html,v 1.22 2005/11/07 20:48:40 melamed Exp