Statistical machine translation (SMT) is a machine translation paradigm where translations are generated on the basis of statistical models whose parameters are derived from the analysis of bilingual text corpora. The main benefits of SMT over traditional (rule-based) paradigms are that the engines produce more appropriate and natural sounding translations. Moreover building an SMT engine requires less manual development work and the technology is not tailored to any specific pair of languages. The success of the MT system development however depends on the availability of a substantial, qualitative parallel corpus.
CrossLang creates custom-built MT engines based on Moses Statistical MT, an EU-funded open source project under the guidance of Dr. Philip Koehn.
The Moses toolkit consists of all the components needed to preprocess data, train the language models and the translation models. It contains tools for tuning these models and automatically evaluating the resulting translations. However, with the steadily increasing demand for high quality translation, it is also becoming a common open source SMT solution for the localisation industry and this is not by chance: Moses works best in scenarios where TM systems work best – large existing translation resources, mainstream language pairs and narrow domains. For many users, custom Moses systems have not only provided better translations in such scenarios compared to widely used SMT tools but combined with post-editing, they have also proven to be more productive than TM alone. As a result, the Moses toolkit has gained a lot of interest from the language industry in the recent years and this interest seems to be spreading with an accelerating rate.
The main benefits of the Open Source direction are:
- Technology ownership - no on-going licence fees
- Custom-built means it will work best for your needs
- No long term dependency on vendors – community development will ensure continuity and support
The main commercial arguments against Open Source are countered by the services we offer around the platform – as well as development we also provide:
- Implementation – including integration with existing technologies
- Documentation and training – both IT and user
- Technical support and maintenance