Neural machine translation method and device based on knowledge graph, equipment and medium

Abstract

The invention provides a neural machine translation method and device based on a knowledge graph, equipment and a medium, and the method comprises the steps: obtaining an original bilingual parallel statement pair, extracting a word and phrase translation pair according to the original bilingual parallel statement pair, and obtaining a corresponding seed entity translation pair; obtaining a source language knowledge graph and a target language knowledge graph, and constructing a corresponding vector space according to the seed entity translation pair, the source language knowledge graph and the target language knowledge graph; when a to-be-translated entity set is obtained, deducing the to-be-translated entity set according to the vector space to obtain a corresponding to-be-translated entity translation pair; and calculating the distance between the seed entity translation pair and the to-be-translated entity translation pair, and obtaining a pseudo bilingual parallel sentence pair containing the to-be-translated entity translation pair according to the distance. According to the method, the knowledge graph is fused into the neural machine translation, and the entity translation accuracy of the neural machine translation is improved by utilizing rich entity knowledge in the knowledge graph.

Description

technical field [0001] The present invention relates to the technical fields of natural language processing and machine translation, and in particular to a neural machine translation method, device, device and medium based on a knowledge map. Background technique [0002] Machine translation is the conversion between different languages ​​by computer, that is, the conversion from source language to target language by computer. Neural machine translation is based on machine translation and realizes automatic conversion from source language to target language through neural network. Although the current neural machine translation is outstanding, there are still deficiencies, especially in the process of entity translation. The current neural machine translation has poor quality of entity translation, which makes the accuracy of entity translation of neural machine translation low. Contents of the invention [0003] The present invention provides a neural machine translatio...

AI Technical Summary

Problems solved by technology

Although the current neural machine translation is outstanding, there are still deficiencies, especially in the process of entity translation

The current neural machine translation has poor quality of entity translation, which makes the accuracy of entity translation of neural machine translation low

Images