17. Semantic issues in machine translation
Kurt Eberle · 2019
There has been research in Machine Translation for half a century. In this article we sketch the role that semantic knowledge and semantic analysis of words, sentences and texts have played in the different architectures that have been suggested for translating texts automatically. We sketch the reasons for making use of semantics in Machine Translation, the problems that arise from using it, the shape tradeoffs could take for minimizing the cost of semantic analysis and maximizing its impact on translation quality and how linguistic analysis and semantic representation could be combined with data-driven approaches and why this should be done.