Machine translation evaluation inside QARLA

In this work we present the fundamentals of the IQMT framework for MT evaluation. IQMT offers a common workbench on which existing evaluation metrics can be utilized. We suggest the IQ measure and test it on the Chinese-toEnglish data from the IWSLT 2004 Evaluation Campaign. We show how the correlation with human assessments at the system level improves substantially for most individual metrics. Moreover, IQMT allows to robustly combine several metrics avoiding scaling problems and metric weightings. Several metric combinations were tried, but correlations did not further improve significantly.