OPERATION METHOD AND APPARATUS FOR NEURAL NETWORK TEXT TRANSLATION MODEL, AND DEVICE AND MEDIUM

摘要

Disclosed are an operation method and apparatus for a neural network text translation model, and an electronic device and a storage medium. The neural network text translation model comprises an encoder layer, an attention mechanism layer and a decoder layer. The method comprises: inputting a source language word sequence into an encoder layer for processing, so as to form an implicit structure vector; controlling an attention mechanism layer to generate a word alignment table; inputting the implicit structure vector and a context vector, during the translation of each word, into a decoder layer for processing, so as to generate a target language word sequence; acquiring unknown characters in the target language word sequence, and determining, according to the word alignment table, a source language word, which corresponds to the unknown characters, in the source language word sequence; translating the source language word to obtain a target language word; and replacing the unknown characters in the target language word sequence with the target language word. Therefore, the amount of unknown characters in a translation result can be reduced or even completely eliminated.