Evaluation of Machine Translation

since 2012

This project was funded in an initial phase (2012 -2013) by EAMT

The aim of this project is to investigate the use of a constraint dependency parser, involving semantic und syntactic information in the evaluation of machine translation systems. We intend to investigate if and o which extent this kind of parser and the proposed metric can detect structural problems in statistical and example based machine translation

The proposed metric will be tested on German as target language. The choice of German is motivated by the availability within the proposer organisation of a constraint dependency parser, but also by the complexity of the German language .We will use source language texts from other langauge families : Romanian (Latin), Bulgarian Polish, Croatian / Slavonic, Greek. For comparison reasons we will involve also English.

Persons involved: Melania Duma (Ph.D work), Cristina Vertan, Walther v. Hahn, Wolfgang Menzel