Patent Machine Translation

We can consider two actors that may be interested on the patents machine translation: a web user (indeed, the web application) and a (European) patent editor. The goal of the former may be to obtain translations of a patent content. Instead, the goal of the latter is to translate the original text of a patent to any (or all) the oficial languages of the European patents (English, French and German). In such case, the patent editor may be want to post edit the translation.

In MOLTO we focus on the first actor. To do so, we develop a hybrid machine translation system able to produce high quality translation of the patent's claims and abstracts. Then, we will propose a pipeline that produces the missing translations of the patents included in our databases. Hence, end-users will be able to obtain the patents text in any of the three languages.