Chapter-level neural machine translation method and system based on routing algorithm

Abstract

The invention discloses a chapter-level neural machine translation method and system based on a routing algorithm, and belongs to the technical field of natural language processing application. According to the method, context information input into a model system is screened through a routing algorithm, some words useful for sentences to be translated are actively selected from the sentences to be translated to form needed chapter-level information, the information is modeled through the model by introducing the routing algorithm, and therefore the chapter-level neural machine translation effect is improved. The system comprises a data processing module, a routing algorithm module, a context encoder module, a source language encoder module, a decoder module and a generator module. Compared with the prior art, the method and system have the advantages that the anterior text and the posterior text are simultaneously used as required chapter-level information, the to-be-translated sentence automatically selects the words in the context as a part of the required information by utilizing the routing algorithm, the chapter-level neural translation model is constructed, the context information is effectively utilized, and the chapter-level machine translation quality is improved.

Description

technical field [0001] The invention relates to a neural machine translation method and system, in particular to a routing algorithm-based text-level neural machine translation method and system, and belongs to the technical field of natural language processing applications. Background technique [0002] Thanks to the development of deep learning, neural machine translation has made great progress in most language pairs. Most of the existing standard neural machine translation methods are aimed at sentence-level translation. Using an end-to-end learning method, an encoder-decoder structure model is constructed, and the source language sentence and its target language sentence are used as model input for training the model and Learn the corresponding relationships. However, when sentence-level machine translation is applied to text-level machine translation, it will lead to chapter-level problems such as ambiguity of reference, inconsistency of translation before and after, ...

AI Technical Summary

Problems solved by technology

Although such methods are straightforward, it is difficult to distinguish useful information from the noise brought by contextual encoding

At the same time, due to the increase in encoding length, it is difficult for the model to model the relationship between sentences

Images