Machine Translation and Post-editing

The process of machine translation and post-editing blends the productivity of automation with the clarity and quality of human labour. Due to the arise of the so-called Neural Machine Translation (NMT) technology one can achieve significant efficiency with even grammatically, syntactically and semantically complex languages as Hungarian or Slavic languages. In order to achieve maximum productivity and cost-efficiency we advise our customers to choose the Clean Data Model according to which only high standard client and customer specific language repositories - translation memories, mono- and bilingual terminology databases - are used for the training of machine translation engines.

When it comes to post-editing machine translations we differentiate between two types of quality levels. Full-post editing results in grammatically, syntactically and semantically impecable target language texts similarly to those completed by professional translators and reviewers. However, it is more time consuming and more costly than Light Post-Editing. Latter is recommended when volumes are high, deadlines are short and the objective is not a fully accurate target text but rather the support of overall comprehension of a foreign language content. In this case post-editors only focus on correcting critical syntactical and semantical errors that largely aggravate comprehension, while leaving minor spelling and grammatical errors intact.

In order to get the desired language quality, preparation of the MTPE project and training of the MT engine is as important as applying professional language experts for post-editing, therefore we are at your disposal already from the project planning phase.