Speaker: Andreas Maletti, researcher at Institute for Natural Language Processing, Stuttgart, Germany 

Abstract: We will cover the basics of extended tree transducers and their application in the area of statistical machine translation. Syntax-based statistical machine translation, in which the translation is based, at least in part, on syntactic trees, uses several tree transducer models. We introduce two extended tree transducers in detail with a strong emphasis on examples. In particular, we show how to obtain such transducers automatically from data. In addition, the basic theorems on the expressive power of these devices and on some relevant properties are presented. We conclude with a recent empirical evaluation of the introduced devices.