\\ Technologies :: Translation :: Reuse

There is a distinction to make between recycling and reuse.

Translation memory uses the source text to find existing translations. The result is a list of candidates. There is never, in a standard TM implementation, an absolute certainty that an exact match will provide the correct translation for the context where it will be placed.

True (safer) reuse can be achieve with other techniques, that belong usually more to the document management category, such as matching based on IDs.

For example:

Version 1 English:
  Switch the printer on. It starts in Save mode.
Version 1 French:
  Allumez l'imprimante. Elle démarre en mode Econnomie.

Version 2 English:
  Switch the copier on. It start in Save mode.
Version 2 French: (second segment is exact match):
  Allumez le copieur. Elle démarre en mode Econnomie.
Version 2 French: (correct):
  Allumez le copieur. Il démarre en mode Econnomie.

Translate the two sentences of version 1, using a TM. Then, translate the two sentences of version 2, using the same TM. If you get an exact match for the second sentence of version 2, the tool is doing recycling, not re-use and your exact matches should be verified if you want to be sure they are correct. If they are not verified you might risk context-related problem like the one shown here.

Some commercial tools are starting to additional tools to provide better support for true reuse, by, for instance, checking the segment before the current one: if it is also an exact match, chances are better that the context is the same. But even these techniques do not always provide perfect solutions.