Site Translator : What is Translation Memory?

What is Translation Memory?

Translation Memory can be defined as a list of source and translated target language value pairs.  It is based on fuzzy logic and is the ultimate solution to the problem of inaccuracy of Machine Translation faced by many users.

Let us try an understand how Machine Translation works.  Machine Translation substitutes words in one language for those in another language.  Understanding sentences and translating is a complex procedure where the meaning of the source text has to be understood in its entirety.  All features of the text must be analyzed, a process which requires in-depth knowledge of both the grammar, semantics, syntax, idioms of the source language, as well as the culture of its speakers.  The same in-depth knowledge is needed to re-encode the meaning in the target language.  The main challenge for Machine Translation is to understand nuances of both the source and the target language.

You will often translate the same text (phrases, words, or sentences) over and over again.  The translation provided by machine translation could be inaccurate.  You would like to be able to edit the translated text and store the result.  Then, the next time you translate the same text, you get the translated text you previously defined for that source text. 

Translation Memory helps you to customize the database of translated phrases to overcome the inherent flaws in the word substitution process on which Machine Translation works.  You can edit the translated values returned by Machine Translation or hire people to translate certain information for your organization and provide your clients with a consistent and accurate translation.  Translation Memory is based on fuzzy logic i.e. it can give results even when the data is not totally accurate, precise and logical, as in case of languages which can never be mathematically defined.  The accuracy of Translation Memory improves with increasing number of inputs and outputs.

The Percentage Match i.e. Percentage accuracy for finding the match of a source text in the Translation Memory is set to 100% by default.  Every source text is automatically searched in the Translation Memory and only if not found  is it sent for Machine Translation.  Machine Translation in the context of Site Translator means that the text to be translated is searched for in a remote language server and then translated.