Machine Translation System and Formats

PangeaMT was born as a machine translation technology based on the 2nd release of Moses. Pangeanic was the first company in the world to implement Moses commercially -as reported by EU FP7 Research Programme Euromatrixplus. It was designed from the very beginning  to serve open standards workflows (inputs and outputs in TMX, xliff). It is now offered as a customization package built for the needs of the following industries:

  • Engineering
  • Automotive
  • Electronics / Computer Hardware and Peripherals
  • Telecomms
  • Computer Software
  • Marketing-Economy-Business
  • Legal-Professional services
  • Life Science / Medicine

PangeaMT’s Machine Translation System and Formats are modular. It is a widely implemented, successful solution serving thousands of words a day in completely private environments and tens of thousands of words for gisting.

With PangeaMT, you can create and add new translation engines with your own material in the language pairs you have purchased and also create language domains of your choice for your own engine. PangeaMT is constantly experimenting and adding more domain-specific engines. thanks to its powerful API, translation is served in milliseconds to the environment of your choice.

One of the main benefits of PangeaMT is that your engines are held in a totally private environment: in your server. If you are looking for cloud solutions, do not hesitate to send an email to our sales team.


The output from PangeaMT is often praised for its efficient handling of tags, unlike other plain-text solutions. This was already reported in 2010 and 2011 at Localization World and TAUS by several clients, like Sybase and Sony Europe. Currently, PangeaMT is fully compatible with older versions of Trados, and with current versions of SDL Studio, MemoQ, MemSource, as well as most formats supported by the Okapi Framework (xliff, InDesign idml, FrameMaker mif, html, all LIbreOffice formats, etc).

When using our translation API for translation, text can be extracted from your website and sent directly to one of engines and returned to your CMS for publication and forwarded to a human post-editor. Contact PangeaMT to find out more how we can set up a plugin and connector to your CMS.

Below follows one description of pre-translation using a standard ttx format which is still very popular in some organizations.

TTX Worflow

How does it work?– Choose the level at which it is preferable to send new content to the SMT system (typically matches under 70%). Export your Project Translation Memory (preferably untranslated segments only). You won’t pay for 100% matches, only for the real new text you need! Your output file will be in tmx (xml format) and should look like this:


– PangeaMT will place a translated segment in the target segment like so:


-Now import the TMX into your translation system. Set a penalization for translator MT!.

– Start post-editing: your translation program will stop at untranslated segments over 70% match, which require little translation and at those translated by MT! for quick post-editing.


Remember PangeaMT has developed using generic and self-generated data following language models which may not be perfectly fitting to your needs. Results may very according to the type of text you provide. We cannot guarantee our raw MT will fit your particular needs or terminology.


As with any post-editing exercises, we recommend you prepare a terminology interface /consistency-checking application. If post-editing at Pangeanic, we will require an approved terminology list to ensure terminology consistency. If you translate over 100.000 words in a year into a specific domain, we recommend you to consider your own SMT development. Please ask our consultancy team for a details, necessary data and a quote.

Start using PangeaMT today!