基于知识的智能机器翻译方法研究

机器翻译近年来取得了快速发展，但离实际应用要求还有一定的差距。本项目以实际应用为目标，重点研究基于知识的智能机器翻译方法，提出一套实用的面向机器翻译的跨语言、多模态语义表示理论和方法；研究融合先验知识的神经机器翻译框架，提高机器翻译的忠实度；设计实现面向机器翻译的在线人机交互学习算法，建立翻译知识的在线学习和动态更新机制；研究利用大范围语境信息提高神经网络机器翻译系统性能的方法；实现高水平、多语言智能机器翻译系统，并得到实际应用。这项研究不仅对于丰富和发展机器翻译理论，推动自然语言处理相关技术的发展，具有重要的理论意义，而且对于提升机器翻译系统的智能化程度，更好地为用户提供个性化翻译服务，具有重要的应用价值。

Despite the rapid development of machine translation in recent years, there is still a long way to go to meet the application requirements. This project aims at practical application, focusing on the study of knowledge-based intelligent machine translation methods. In this project, we are planning to propose some practical theories on cross-lingual multi-modal semantic representation and methods for machine translation, establish a neural machine translation framework which integrates prior knowledge to improve translation adequacy, and design an online human-computer interaction learning algorithm for machine translation, which can learn and dynamically update translation knowledge online. Finally, we will implement a high-level and multilingual intelligent machine translation system and apply it in practice. This project has a great significance for enriching and developing the theory of machine translation and promoting the related technology of natural language processing. It is also of great significance to enhance the intelligence of machine translation systems and provide the better personalized service.