Translation method and device based on neural network

Abstract

The embodiment of the invention discloses a translation method and device based on a neural network. The method comprises the following steps that: obtaining the initial translation of a sentence to be translated, wherein the initial translation carries unlisted words; splitting the unlisted words in the initial translation into characters, and inputting a character sequence formed by the characters obtained by splitting into a first multilayer neural network; through the first multilayer neural network, obtaining the character vector of each character in the character sequence, and inputtingall character vectors of the character sequence into a second multilayer neural network; using the second multilayer neural network and a preset common word database to code all character vectors to obtain a semantic vector; and inputting the semantic vector into a third multilayer neural network, decoding the semantic vector through the third multilayer neural network, and combining with the initial translation of the sentence to be translated to determine the final translation of the sentence to be translated. The method has the advantages that the translation operability of the unlisted words can be improved, the translation cost of machine translation is lowered, and the translation quality of the machine translation is improved.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a neural network-based translation method and device. Background technique [0002] Currently, in the process of statistical machine translation, since the translation model of statistical machine translation is automatically learned from the training data, for words that have not appeared in the corpus trained by the translation model, the translation model cannot generate the translation corresponding to the word, thus appearing The phenomenon of unregistered words. Wherein, the above-mentioned unregistered words are words that have not appeared in the training corpus of the translation model, and the translation result obtained by the translation model is generally output as it is or as "unknown (UNK)". In statistical machine translation, especially in cross-domain machine translation (for example, the translation model trained in the news field is used for translation...

AI Technical Summary

Problems solved by technology

The update frequency of new words in network text data is high, the operability of updating and maintaining dictionaries in time is poor, and the implementation is difficult, which makes the realization of machine translation with the help of dictionaries difficult and costly

Images