With the rise of Machine Translation technologies and solution, the demand for Post Editing Machine Translation Services has increased considerably. We have been providing PEMT services into multiple languages and multiple fields since 2007. It means that we have been working in PEMT since the early staged of Machine Translation when a very few No. of languages were supported and when the MT engines were very weak compared to what we have today.
Now, with the breakthrough advances in Machine Translation technologies and methodologies, it is capable of producing higher quality for most of the languages in a very short time. These breakthrough advances led to an increase in demand of human-MT-post-editing services especially for multinational companies and organizations with huge amounts of content/data processed on a daily basis as well as smaller companies/organization seeking to reduce their content translation/localization costs.
Currently, there are four types of Machine Translation Engines:
- Statistical Machine Translation (SMT)
- Rule-based Machine Translation (RBMT)
- Hybrid Machine Translation
- Neural Machine Translation.
Types of Post-Editing
Full Post Editing (FullPE)
This type of PE is supposed to produce human-like quality and it can take a longer time. The resulting text should sound very natural and free of translation or linguistic errors. The rate of this type of post editing is normally higher as it requires much more effort than Light Post Editing (LPE). This type of PE is usually used when the client has is handling sensitive data, has enough time, budget and requires a high quality output. The Post Editor in this model is responsible for all aspects including accuracy of meaning, language, style and tone of speech.
Why then do full PE clients need Machine Translation in first place if they are going to pay almost the same cost and spend almost the same time? From our expertise, they use MT + full PE to train their MT engines and prepare it for light PE in a later phase, i.e. it’s a temporary phase to reduce the cost and time later on.
Light Post Editing
This is also named Rapid Post Editing. In Light PE the post editor is asked to improve the quality of the resulting machine translated text so that it gives the reader/audience a rough idea about the source content while making sure that the main ideas of the source are clear in the target language.
Light PE is usually used when:
- Client has huge amounts of data that needs to be translated in a very short time and their priority is having this data translated quickly but it does not have to be perfect. They just need the target reader/audience to understand the basic ides of the source text while it is not necessary to refine the style and other linguistic issues. They usually need to translate their new or blog content or social media posts once they are published and can’t wait for a full cycle of full PE.
- Client already has a well-trained MT engine and trusts that it will produce a good quality when applied to his source content. These are mainly companies that have long-term investments in the MT field and has already passed the full PE phase. An example for this client is the hotel-booking websites who have already a huge corpse of already-translated and revised content that’s used by the machine-translation engine to process new hotels/resorts/properties. After that they need a light-PE to fine tune the output.
- Client has low budget for translation and just need to give rough idea about their product/services in new markets.
