The GenPar Toolkit for Research on Generalized Parsing
(including Machine Translation by Parsing)

This toolkit provides an architecture, a design, and an implementation of an integrated system for statistical machine translation by parsing. It was built to be flexible, extensible, and easy to understand. It is

dynamically configurable in many ways
accompanied by 3 kinds of documentation: user, system, and design
written in object-oriented C++
freely downloadable below

License and Citation

GenPar is available under the terms of the GNU General Public License (GPL) version 2.0 or later.

After pilot work at NYU, GenPar was developed by the Statistical Machine Translation by Parsing team during the John Hopkins University Center for Language and Speech Processing 2005 summer workshop. That team released GenPar-1.1 and its final report (below) in November 2005.

The following people contributed to the effort of getting from GenPar-1.1 to the latest version (in alphabetical order): Andrew Chizhik, Markus Dreyer, Daniel Galron, Tz-Huan Huang, Christo Kirov, Dan Melamed, Chris Pike, Thomas Torsney-Weir, Joseph Turian, Sonjia Waxmonsky, and Ben Wellington.

If you use GenPar in your research, we would appreciate the following citation in any resulting publications:
A. Burbank, M. Carpuat, S. Clark, M. Dreyer, P. Fox, D. Groves, K. Hall, M. Hearne, D. Melamed, Y. Shen, A. Way, B. Wellington, and D. Wu (2005) "Final Report of the 2005 Language Engineering Workshop on Statistical Machine Translation by Parsing", http://www.clsp.jhu.edu/ws2005/groups/statistical/documents/finalreport.pdf .

Downloads

After unpacking the tarball, look in the file README.1st for further instructions.

We expect this software to evolve rapidly, so we strongly recommend that users sign up for the very low-volume genpar-announce mailing list, to be notified of bug fixes and new features.

The latest version of this software is GenPar-1.3.4:

Binaries:
Warning: These binaries are statically linked, to avoid library problems, but this makes them LARGE.
genpar-1.3.4.Linux.tar.gz for Linux
genpar-1.3.4.SunOS.tar.gz for SunOS (Solaris)
Source:
genpar-1.3.4.src.tar.gz

Click here to see the list of previous releases.

Add-Ons

GenPar users might also be interested in our Multitree Viewer (MTV).
Language-specific resources are here.

FAQ List

is here.

Mailing Lists

We expect GenPar to evolve rapidly, and to release improved versions. If you'd like to be notified of major bug fixes and new features, please subscribe to the very low-volume announcements mailing list via this web form.

User Support

Please send bug reports, feature requests, and other suggestions for improvement to genpar-devel AT cs DOT nyu DOT edu. Please note that this list is moderated, and only pertinent messages will be posted.

How to join the project

GenPar is an open source project and we welcome volunteer contributors. Please look here for information on how to do that. For more information, you are welcome to contact the project leader, Dan Melamed, by sending email to lastname AT cs DOT nyu DOT edu.

Acknowledgments

In addition to the people listed at the top of this page, the following people made significant contributions to this project. Many thanks to them, and apologies to those whom we accidentally omitted:

Josh Rosenblum, Noah Smith, and Svetlana Stenchikova, for the genesis of the "common" containers
Wei Wang, for initial work on GenPar and its documentation
Fred Jelinek, for funding our team at WS'05
Victoria Fossum, for help with simplifying data structures
Ali Argyle, for help with organizing the 2nd planning meeting
Mona Diab and Nizar Habash, for help with Arabic tools
Dan Bikel, for help with his parser, and for giving us a special license
Abhishek Arun and Frank Keller, for help with the French treebank
Charles Schafer, for help with running GIZA++ baselines

The GenPar Toolkit for Research on Generalized Parsing (including Machine Translation by Parsing)