Document home

Difference between termbase, translation memory, existing translations, and machine translator

Soluling supports four different translation leverage technologies. They are termbase, translation memory, existing translations, and machine translator. Even at first sight, they might seem very similar, but actually, they are very different and are meant for different purposes.

Termbase

Termbase or glossary typically is the most accurate of these four technologies. It is like an old fashioned dictionary book. You can easily check the meaning, spelling, or translation of a word. Termbases are used to make translating words easier and consistent. Consistency is very important in translation work. It is essential that the same term is always translated in the same way. Using termbases makes this easier for you. Soluling supports several different termbase and glossary formats. A typical termbase is a local database, server database, or service in the cloud. A typical glossary is a TBX, text, or Excel file.

Translation memory

Translation memory is used to remove a need to translate the same sentence more than once. Translation memory saves your previous translations, and if you meet the same sentence (or a very similar one), translation memory brings the existing translation automatically by making the translation job faster. Soluling uses fuzzy matching when searching for matches from translation memories. Translation memory is stored in a local database or service in the cloud.

Storage Importing existing translations

If you have existing localized files, import them first. In most cases, when you create a new project or add a new source into the existing project, Soluling finds the existing localized files and can prompts if you accept. You might also have existing translations in translations storage files or databases. Translation storage is typically a file (e.g., TMX or Excel) or database that contains the translations. Translation storage can contain single words or terms just like a glossary and also whole sentences like a translation memory. Soluling supports several different translation storage formats. You can import translations from existing translation storage into Soluling projects or translation memories.

Disks Machine translator

A machine translator, on the other hand, is more inaccurate. In many cases, translations provided by a machine translator are not good enough for final use. However, a machine translator is very useful if you need to make some prototype translations quickly. The translation done by machine translation is much much faster and cheaper than real translations. In most cases, machine translation can be used as a start point for actual human translations if machine translation is applied first, and then the result translations edited manually. This is called post-editing. Machine translation has taken giant steps in the past years, thanks to neural computing and machine learning. Soluling has interfaces to the best machine translation engines such as Amazon, DeepL, Google, IBM, Microsoft, and Yandex - all supporting state of the art neural machine translation.

Translation storage and existing translations are typically used in batch mode, where you translate a large number of rows in a single run. Glossaries are always used interactively on the selected rows and words. Translation memories and machine translation can either be used interactively or in the batch mode. All the above technologies are useful and are used for different purposes. This is why Soluling supports them all.

In addition to the above technologies, you can use Cloud Translation Services, for example, to get high-quality human translations for your projects.