The neural machine translation networks and their algorithms have become a commodity. In a time of need when translations must be done as quickly as possible, it is often necessary to use these neural machine translation systems. But you have to choose the right system, which protects the data to be translated and guarantees good output. If the machine translations are good to begin with, we can further improve on quality over the traditional methods of translation, if subsequently a post-editor checks the texts and corrects them according to clear editing rules.
The role of the translator is changing. Post-editing is the new profession. A post-editor does not start with a translation from scratch, but he has to correct machine translations.
To do this job, post-editors have a guideline that changes from job to job.
It is divided into light-post-editing and full-post-editing. The two methods are self-explanatory - there are differences between the two methods.
Only the most necessary parts are to be changed. Even if the text does not flow well and you notice that it is an automatic translation, the content must be right. This working method is recommended for documents for internal use.
The text is not completely rewritten, but the post-editor pays attention to the company's corporate identity, terminology and style. This working method is used for external documents. The quality of the final text often exceeds the quality of a traditional translation done by a translator, as the post-editor pays a lot of attention to terminology and corporate identity and quality requirements defined by the client at the other end.
In order to have a good machine translation you will need not only the right system, you will also need a well written source text. Many customers ask what to do next. The text must be written according to style and grammar rules. The segmentation must be correct and the sentences must be clear. For example, pronouns or idioms should be avoided.
The team develops a guide on style, which helps to rewrite already written documents and help to make the new documents automatically translatable.