opentag.com
\\ Technologies :: Translation :: Machine Translation

Machine translation is the technology that allows the translation of text automatically.

Good results with machine translation depend in a large part on how good are the lexicons from which the translation will be created (for rule-based MT), or on how much training data you have on a given topic and their quality (for statistical-based MT). Using an MT system for localization can significantly lower costs, but it is a long term investment.

Machine translation can also benefit a lot from controlled language, a form of development of the original document where specific rules are enforced to create a text that will be easier to automatically translate. This entails for example to use simple sentences and a selected set of terms. Some tools provide a way to verify that these rules are respected while creating the documents.

Machine translation can be also used in combination with translation memory systems: the MT tool provides fuzzy matches to the translation tool, allowing the user to easily accept, modify or reject the proposed text. This enable the translator to take advantage of machine translation without having to change the way he or she is working.

Because lexicons are an important asset, being able to port them from one MT system to another is critical. The standard for exchanging lexicons is the Open Lxicon Interchange Format (OLIF).

Some translation systems: