NEW: Across Bundle for Machine Translation

The Concept

Across comprises a translation memory (TM) that stores translations from the past and makes them available for reuse in new translations. Identical recurring sentences are automatically pre-translated, and similar sentences are proposed to the translator for adoption and post-processing.

In contrast, machine translation "learns" how to translate from one language to another by analyzing millions of translated documents. Language Weaver, whose products boost the overall productivity in collaboration with translators and TMs, is one of the leading providers of statistics-supported machine translation.

 

Language Weaver Integration

 

Accordingly, Across has merged all three approaches – human translation, TM, and machine translation – in an integrated solution. If machine translation is defined as an additional workflow step, the procedure is as follows:

  • Each text segment is first analyzed against the Across translation memory.
  • If no matching entries are found, a Language Weaver machine translation will be inserted.
  • Subsequently, the remaining workflow steps are processed, i.e. post-processing by a translator, separate revision, and release.

The Product

The quality of the machine translation greatly depends on the content and volume of the documents with which the system was "trained".

From within Across Language Server, Language Weaver machine translation can be used

  • as a customer-specific trained system for various contents and language combinations,
  • as a standard system for contents from the mechanical engineering/ automotive industries in the language pairs German/English or English/German.

The machine translation bundle of Across represents an inexpensive gateway to the integrated use of automatic translation with Language Weaver. It contains

  • the Across Language Server,
  • the Language Weaver system,
  • the pre-filling of the defined standard system.

The standard system can be customized by adding further contents and/or language pairs.