CAT (Computer Aided Translation) is a term used to classify a wide variety of software tools that are employed to assist the translator in his work, either in consistency, speed, or simply workload. Some definitions include machine translation tools in the CAT category, while others consider CAT to be exactly equivalent to TM.
MAHT (Machine-Assisted Human Translation) can be considered to be either equivalent to or a subset of CAT, depending on the definition of the latter term. MAHT tools are those that do not translate automatically, but assist the translator in his work. All the different definitions of MAHT include Translation Memory tools, and most also consider terminology management tools to be part of the class.
TM (Translation Memory) tools are programs that store previous translations and, upon translating new texts, offer suggestions to the translator, based on material stored in a memory database.
MT (Machine Translation, often called Automatic Translation) systems are those that use linguistic information about the languages being translated to automatically generate translations.
EBMT (Example-Based Machine Translation) is a relatively new technology which aims to combine both TM and MT paradigms by reusing previous translations and applying various degrees of? linguistic knowledge to convert fuzzy matches into exact ones. However, some early definitions of EBMT refer to what is now known as TM, and often exclude the concept of fuzzy matches.
In the context of TM, a memory database (referred to in Déjà Vu X2 as a translation memory) is the file or database where a tool stores previous translations, usually in aligned pairs of source and target language sentences.
Terminology management tools store information in terminology databases (Termbases in Déjà Vu X2). Some TM systems include terminology management databases, and most EBMT tools, such as Déjà Vu, use terminological information to improve the quality of fuzzy matches.
When a TM tool searches for a sentence in a memory database, a stored source sentence that is exactly equal (often ignoring any formatting information that may be stored with it) to the sentence being sought is considered an exact match.
In a TM system, a fuzzy match is a source sentence that only partially matches the sentence being sought.