Principal investigator (ÚFAL): 
Provider: 
Grant id: 
976518
Duration: 
2018 - 2020

Neural machine translation - as a fairly new field of computational linguistics
- recently began to outperform all of the existing techniques of automatic
machine translation. This approach achieves very good results even in the task
of open-domain translation. Most of the existing work on neural machine
translation does not make use of any linguistic information, though.
In this project, we will explore the potential of the linguistic knowledge,
which is accessible either as a part of annotated corpora, or obtainable
automatically using automatic NLP tools. We will employ various forms of
linguistic annotation, such as morphological tags or dependency trees. We
assume that adding the linguistic knowledge will enhance the adequacy of the
neural machine translation system outputs, which will lead to overall better
translation quality.