Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
TRANSLATION METHOD, APPARATUS AND DEVICE BASED ON MACHINE TRANSLATION MODEL, AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2020/253669
Kind Code:
A1
Abstract:
Provided are a translation method, apparatus and device based on a machine translation model, and a storage medium. The method comprises: encoding an ith source word embedding of a source sentence into an intermediate vector; decoding the intermediate vector in conjunction with the ith source word embedding to obtain a decoded intermediate vector; fusing the decoded intermediate vector with an (i-1)th target word embedding of a target sentence to obtain a fused intermediate vector; decoding the fused intermediate vector to obtain a decoded word vector; and performing probability prediction on an ith target word according to the decoded word vector, and outputting a prediction result. According to the method, a previous target word embedding is added in a decoding process to serve as input data so as to capture a dependency relationship between words, such that the accuracy of generation of a target sentence is improved, mistakes such as over-translation and under-translation are reduced, and the generated target sentence is more fluent.

Inventors:
SHAO CHENZE (CN)
ZHANG JINCHAO (CN)
MENG FANDONG (CN)
FENG YANG (CN)
ZHOU JIE (CN)
Application Number:
PCT/CN2020/096283
Publication Date:
December 24, 2020
Filing Date:
June 16, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
TENCENT TECH SHENZHEN CO LTD (CN)
International Classes:
G06F17/00
Foreign References:
CN110457713A2019-11-15
CN109271646A2019-01-25
CN109508462A2019-03-22
US20190130249A12019-05-02
Other References:
CHENZE SHAO, YANG FENG, JINCHAO ZHANG, FANDONG MENG, XILIN CHEN, JIE ZHOU: "Retrieving Sequential Information for Non-Autoregressive Neural Machine Translation", ARXIV.ORG, 22 June 2019 (2019-06-22), pages 1 - 12, XP081385799
ZHANG JINCHAO;AISHAN WUMAIER;MAIHEMUTI MAIMAITI;LIU QUN: "A Large-scale Uyghur-Chinese Neural Machine Translation Model Based on Multiple Encoders and Decoders", JOURNAL OF CHINESE INFORMATION PROCESSING, vol. 32, no. 9, 15 September 2018 (2018-09-15), pages 20 - 27, XP055767286, ISSN: 1003-0077
Attorney, Agent or Firm:
BEIJING SAN GAO YONG XIN INTELLECTUAL PROPERTY AGENCY CO., LTD. (CN)
Download PDF: