Why we Chose PMNT

For over forty years, automated translation engines have evolved continuously by integrating technologies of their time, from symbolic systems (based on rules) to statistical models. Today, the maturity of Big Data technology, coupled with the progress made in Artificial Intelligence , bring the power and sophistication necessary to propel the translation into the new stage of Deep Neural Network, the artificial neurons and the self-learning machine. We have chosen Systran’ Pure Neural Machine Translation (PNMT) system not only because Systran is the world-leader in machine translation, but also because after testing a few software, we believe that PNMT is by far the best tool on the market.


How and when does ON TARGET use Pure Neural Machine Translation?

ON TARGET believes that machine translation is best used for technical documents or documents presenting a high-level of repetitions, a catalog of auto parts for example. We could also employ it when a client’s budget is very limited and the translation requested is not strategically important. We call these documents “For comprehension only”. However, as a rule, we will never use it for marketing documents.

At all times, we do a full post-editing to generate a translation that it stylistically appropriate and linguistically correct. This type of editing often involves the same checks as the traditional translate-edit-proof (TEP) process. It includes checking attributes such as syntax, grammar, style, fluency, and terminology.

What can PNMT do for you?
-Save time
-Lower your costs


Here is what typical a Systran PNMT workflow entails:

1. The translator selects the language pairs needed and initiates the PNMT.
2. The Systran system returns the results in a few minutes and open-up the editing panel (see the image below) for the translator.
3. The translator or editor chooses the vocabulary and edits the phrase as needed.
4. Systran PNMT learns from the editor’s changes and modifies the rest of the text accordingly, storing the results in the translation memory.
5. Once the translator has reviewed the whole Systran output, he or she generates a document in the target language.
6. The translation is grammar checked.
7. The finalized translation is sent to the client.

Click here to see Systran’s own marketing brochure translated from French to English with PNMT.