Moses: Open Source Toolkit for Statistical Machine Translation

The paper describes Moses, an open-source toolkit for statistical machine translation. Its novel contributions are threefold: support for linguistically motivated factors, confusion network decoding, and efficient data formats for translation models and language models. These features extend the standard statistical machine translation decoder with richer linguistic representation and more flexible handling of ambiguous input.

In addition to the SMT decoder itself, the toolkit includes a wide variety of tools for training, tuning, and applying the system to many translation tasks. By bundling the decoder together with this end-to-end tooling as open source, Moses gave researchers and practitioners a complete, reusable pipeline covering the full workflow of building and running statistical machine translation systems.