Translation method and translation device based on neural network model

Abstract

The embodiment of the invention provides a translation method and a translation device based on a neural network model, wherein the translation method based on the neural network model comprises the following steps: obtaining a statement of a source language; coding the statement of the source language to obtain a vector sequence; predicting corresponding candidate words in a target language word by word based on the vector sequence; and generating a statement of the target language according to the candidate words obtained by prediction. The translation method and the translation device based on the neural network model in the embodiment of the invention are capable of performing translation in combination with a variety of translation characteristics; and thus, the translation quality, the fluency and the readability of a translation result are improved.

Description

Technical field [0001] The present invention relates to the technical field of machine translation, in particular to a translation method and device based on a neural network model. Background technique [0002] In recent years, Recurrent Neural Network (RNN) technology has been widely used in the field of machine translation. Compared with traditional statistical machine translation systems, machine translation systems based on recurrent neural networks can make full use of global semantic information, and the translation quality is significantly improved. [0003] However, machine translation technology based on recurrent neural networks also has obvious shortcomings: [0004] 1. The vocabulary is restricted. [0005] Generally, the RNN translation model can only use a limited number of word vocabularies (usually within 30,000 words), resulting in out-of-vocabulary (OOV) words that cannot be translated. [0006] 2. The RNN translation model only supports bilingual sentence pairs for...

AI Technical Summary

Problems solved by technology

[0005] Usually, the RNN translation model can only use a limited number of word vocabulary (usually less than 30,000 words), resulting in words outside the vocabulary (Out-of-vocabulary, OOV) cannot be translated

[0006] 2. The RNN translation model only supports bilingual sentence pairs for training, and it is difficult to use the target language monolingual corpus that can effectively improve the fluency of translation results for training

[0007] 3. Unable to add or enrich more features

However, these features that can improve translation quality cannot be directly used by RNN translation models

[0009] 4. The phenomenon of missing words is serious, and it is easy to generate short translation results, which affects the readability of the translation results

Images