The main objective of the project is to design and implement a new linguistic-stochastic method for machine translation, which we call Compositional Machine Translation (CompMT). The current strategies in statistical and neural machine traslation (SMT/NMT) rely on the use of parallel and aligned corpora, as well as non-compositional phrase segmentation. By contrast, CompMT is based on:
The use of non-parallel bilingual corpora and compositional distributional semantics.
Our specific objective is to implement a Web application to translate ambiguous English sentences with phrasal verbs into Spanish.