首页|Research Results from Northeastern University Update Understanding of Machine Tr anslation (ZeUS: An Unified Training Framework for Constrained Neural Machine Tr anslation)
Research Results from Northeastern University Update Understanding of Machine Tr anslation (ZeUS: An Unified Training Framework for Constrained Neural Machine Tr anslation)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Current study results on machine trans lation have been published. According to news reporting from Shenyang, People's Republic of China, by NewsRx journalists, research stated, "Unlike general trans lation, constrained translation necessitates the proper use of predefined restri ctions, such as specific terminologies and entities, during the translation proc ess." Our news journalists obtained a quote from the research from Northeastern Univer sity: "However, current neural machine translation (NMT) models exhibit proficie nt performance solely in the domains of general translation or constrained trans lation. In this work, the author introduces the zero-shot unified constrained tr anslation training framework, which adopts a novel approach of transforming cons traints into textual explanations, thereby harmonizing the tasks of constrained translation with general translation. Furthermore, the author discovers the pivo tal role of constructing synthetic data for domain-specific constrained translat ion in enhancing the model's performance on constrained translation tasks. To th is end, the author utilizes large language models (LLMs) to generate domain-spec ific synthetic data for constrained translation. Experiments across four dataset s and four translation directions, incorporating both general and constrained tr anslations, demonstrate that models trained with the proposed framework and synt hetic data achieve superior translation quality and constraint satisfaction rate s, surpassing several baseline models in both general and contrained translation ."
Northeastern UniversityShenyangPeopl e's Republic of ChinaAsiaEmerging TechnologiesMachine LearningMachine Tr anslation
NETL
NSTL
