On the Role of Syntactic Trees in Machine Translation

old_uid11831
titleOn the Role of Syntactic Trees in Machine Translation
start_date2012/11/16
schedule11h-12h30
onlineno
summaryOver the past years, major effort was spent on incorporating monolingual syntax into statistical machine translation with mixed results. One difficulty in doing so often seems that translation equivalence does not necessarily conform to monolingual syntactic constituency structure, as assumed by traditional compositional semantics. In this talk I will reflect on syntactic structure aiming to show that the information represented in a syntactic tree can be represented in alternative ways. By representing this information in terms of translation equivalence, syntactic information can be brought into statistical MT models effectively.  In the talk I will give a bird's-eye view of three recent models developed together with my co-workers and touch upon some relevant aspects of the interplay between syntax and SMT. Subsequently, I will give a brief overview of the problem of how to represent the translation mapping hierarchically in order to capture the relations of translation equivalence adequately. I will also briefly discuss how to use this new hierarchical representation to pave the way for learning compositional statistical synchronous grammars from word-aligned parallel corpora.
responsiblesCandito