Machine Translation Based on Syntactic Structure

Abstract

Machine translation refers to the process of translating source language sentences into semantically equivalent target language sentences through computers, and is an important research direction in the field of natural language processing. Neural machine translation only needs to use neural networks to achieve end-to-end translation from source language to target language, which has become the mainstream direction of machine translation research. There are also some problems such as fluent but not faithful translation, difficulty in processing rare words, poor performance of low-resource languages, poor cross-domain adaptability, and low utilization of prior knowledge. Inspired by the research of statistical machine translation, it has become a hot topic in the field of neural machine translation to integrate linguistic information into the neural machine translation model and utilize the existing linguistic knowledge to alleviate the inherent difficulties faced by neural machine translation and improve the translation quality. According to the classification system of grammatical units, researches in this field can be divided into three categories: neural machine translation (NMT) integrating word structure information, neural machine translation integrating phrase structure information and neural machine translation integrating syntactic structure information. Current researches also focus on these three aspects. Firstly, the challenges and causes of neural machine translation are described, and then the research status and main results of neural machine translation based on language knowledge are introduced. At last, the questions in existing research are summarized, and the future research direction is prospected.