This repository contains scripts and an example configuration files used for the Tartu Neural MT submission (C-3MA) for the shared translation task at the 2017 Workshops on Statistical Machine Translation http://www.statmt.org/wmt17/

• NMT-combination - scripts for combining output of NMT systems
• NMT-config - example configuration files for training NMT systems with Neural Monkey and Nematus
• named-entity-translation - scripts for prepairing parallel corpora with named entity tags
• filtering - scripts for filtering parallel text files
• scoring - scripts for scoring translations with language models
• post-processing - scripts for post-processing output translations
• shuffle_parallel.sh - shuffle two files of parallel sentences

The title is derrived from the authors first names (Chantal, and three Ma's - Maksym, Mark, Matiss). It's also a reference to C-3PO - the translation droid from Star Wars.

If you use this tool, please cite the following paper:

Matīss Rikters, Chantal Amrhein, Maksym Del, Mark Fishel (2017). "C-3MA: Tartu-Riga-Zurich Translation Systems for WMT17." In Proceedings of the Second Conference on Machine Translation (WMT17) (2017).

@inproceedings{Rikters-etal:2017:WMT,
	author = {Rikters, Matīss and Amrhein, Chantal and Del, Maksym and Fishel, Mark},
	booktitle = {Proceedings of the Second Conference on Machine Translation},
	title = {{C-3MA: Tartu-Riga-Zurich Translation Systems for WMT17}},
	publisher = {Association for Computational Linguistics},
	address={Copenhagen, Denmark},
	year = {2017}
}