Machine Translation Post-Editing (MTPE)

 

    Machine translation (MT), which is defined as the automated processing by specialized computer software programs know as translation engines, is considered by many to be the future, and also a big part of the present, of translation industry. Translation engines are the software programs that is well-trained and fed by expert linguists, in order to simulate the natural process of human translation, benefiting from the most recent achievements in artificial intelligence (AI) and natural language processing (NLP). There are many types of machine translation, like Statistical Machine Translation (SMT), Rule-Based Machine Translation (RBMT), and Neural Machine Translation (NMT). Each type is based on a different methodology, and thus has its own pros and cons, compared to other types.

The simulation process is based on specific algorithms developed by software engineers in a way that enables translation engines to analyze the context by relating each words to the other words in the same sentence, and each sentence to the other sentences in the whole text. By doing so, the translation engine choses the best option to translate each word and sentence, and generate the translated text in the target language.

Machine translation is widely considered to be the best choice when processing huge volumes of text. The process has proved to be very beneficial, in terms of cost and time saving. Not only that, but also over the course of many years of scientific research and software development works, the overall resulting quality of the translated texts has witnessed remarkable leaps and significantly improved. Yet, and in spite of all such great advancements, the artificial intelligence still cannot, and probably would never be able according to many, to replace human minds. So, machine translation is typically used with texts with low visibility or texts that do not require optimum level of quality, like general, non-sensitive topics, or internal training for employees.

In this context comes the importance of machine translation post-editing (MTPE). In this process, machine-translated texts are edited by human translators and editors, in order to correct the errors that are found in the resulting translation, so that the text sounds flowing and natural. There are two types of machine translation post-editing (MTPE). Each type has different scope of editing and is chosen according to the objective and targeted use of the machine-translated texts. These two types are:

1- Light Machine Translation Post-editing (Light MTPE):
– The objective of this type of editing is only to make the machine-translated text just understandable. The only focus is on delivering the main idea of the source text, without paying much attention to minor errors like grammar or punctuation, or presenting the text in an elegant style.

2- Full Machine Translation Post-editing (Full MTPE):
– In this type, the machine-translated text is fully edited, in terms of grammar, punctuation, readability, and maybe even doing stylistic changes. The aim of this type of editing is to generate a translation that sounds just like natural human translation.

  Ultra Localize has handled many huge-volume machine translation post-editing (MTPE) projects, benefiting from the deep experience of its experienced translators and editors, while exactly following the accurate instructions of each project to fulfill the different needs of each client.