Project hosted
at
Matxin is one of the two open-source machine translation engines being developed inside project OpenTrad; Matxin is designed to translate between different kind of languages (see Documentation below). The OpenTrad consortium is led by Eleka Ingeniaritza Linguistikoa.
Linguistic information for the translation from Spanish and Basque (es-eu) is included.
The project ("Open Source Machine Translation for the Romance languages of Spain") is funded by the Ministry of Industry, Tourism and Commerce of Spain trough PROFIT grant number FIT340101-2004-3 and FIT340001-2005-2.
The Matxin architecture is being developed by the Ixa research group at the University of the Basque Country. Linguistic data for translating from Spanish into Basque using Matxin are being developed by Ixa and Elhuyar R&D.
Matxin is released through the SourceForge page sourceforge.net/projects/matxin/. It contains three types of information:
programs for transfer and generation
data for transfer and generation for the pair es-eu
documentation
Two additional package are necessary for building the whole system:
the lttoolbox package from Apertium (sf.net/projects/apertium/) for the programs
FreeLing from http://garraf.epsevg.upc.es/freeling for the analysis of Spanish texts
Before installing Matxin, you will be able to evaluate it online in www.opentrad.com.
Documentation in Spanish (detailed but still incomplete) is available through the sf.net/projects/matxin/ project page.
We have published a conference paper describing the architecture and its motivation:
Alegria I., Díaz de Ilarraza A., Labaka G., Lersundi M., Mayor A., Sarasola K., M. Forcada, S. Ortiz, L. Padró. 2005. An Open Architecture for Transfer-based Machine Translation between Spanish and Basque. MT Summit. A workshop at Machine Translation Summit X, Phuket, Thailand 2005).