Graph transductions in the field of automatic translation of natural languages

Abstract

First, why do we need automatic language translation at all? There are several reasons. Mainly, there is a growing need for international information exchange. And a lot of the texts to be translated are quite stupid, like service manuals or material lists. And the main idea of automatic translation is that those boring tasks should be done by a machine. Especially for such texts, the dictionary look-up is a main part of the work for a human translator and this can be done already now very efficiently by a machine. Furthermore, the restricted contexts lead to a relatively low ambiguity of the words and syntactic constructions and if one does not have too high expectations concerning the quality of the translation, it should be possible to achieve an automatic translation system. This topic too, is a big challenge for the research in this field and it is still an open question, where the limits of automatic translation will be. In the field of linguistic data processing, there are a lot of interesting problems which can be solved (only) by a machine and the automatic translation can be a promoter for these topics and the tools developed there will be fruitful for the research involved.