Exploiting Context in Multilingual Understanding and Generation. Automatic translation technologies produce incoherent and incorrect outputs in critical areas, such as health, finance, and law. This is due to translating sentences independently, without regard to the global extra-sentential context and rich linguistic structures inherent in the wider document context. This project aims to exploit global linguistic structures, capitalising on recent advances in deep neural networks, in order to g ....Exploiting Context in Multilingual Understanding and Generation. Automatic translation technologies produce incoherent and incorrect outputs in critical areas, such as health, finance, and law. This is due to translating sentences independently, without regard to the global extra-sentential context and rich linguistic structures inherent in the wider document context. This project aims to exploit global linguistic structures, capitalising on recent advances in deep neural networks, in order to generate coherent and faithful text. Expected outcome include next-generation computational technologies for language understanding and generation. This should significantly benefit document-based language technologies and increase their applications in a range of cultural, industrial, and health settings.Read moreRead less
Adaptive Context-Dependent Machine Translation for Heterogeneous Text. While automatic machine translation technologies are undoubtedly useful to a wide range of users, they often produce incoherent outputs for many types of input, for example, medical, literature, or even conversational text. This project will develop new adaptive machine translation systems to handle many domains and text styles, including heterogeneous mixed-domain inputs. It will develop multi-task machine learning methods f ....Adaptive Context-Dependent Machine Translation for Heterogeneous Text. While automatic machine translation technologies are undoubtedly useful to a wide range of users, they often produce incoherent outputs for many types of input, for example, medical, literature, or even conversational text. This project will develop new adaptive machine translation systems to handle many domains and text styles, including heterogeneous mixed-domain inputs. It will develop multi-task machine learning methods for training collections of domain-specific translation systems while leveraging correlations between domains. This approach will reduce the big data requirements of current translation systems, and improve translation quality across a wide range of different language pairs and application domains.Read moreRead less