MACHINE TRANSLATION SYSTEM BASED ON CAPSULE NEURAL NETWORK AND INFORMATION DATA PROCESSING TERMINAL

摘要

A machine translation system based on a capsule neural network and an information data processing terminal, for use in the technical field of computer software. A machine translation method based on a capsule neural network, comprising: mapping words to a high-dimensional vector space to obtain word vectors, obtaining word vectors combining additional information in a corpus, and also combining position information of the words (S101); receiving context vectors, obtaining word vectors to the target language, and obtaining a target language vocabulary by means of a dictionary (S102); In the present solution, by means of changing the internal structure of the neural network, additional information in the corpus can be learned by means of a word embedding layer after improvement of the capsule layer, such that the corresponding words in the source text and the translation are closer, and the distance of words having similar additional information in the source text and the translation in the word vector space is closer.