Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
TRANSLATION PROCESSING METHOD, TRAINING METHOD FOR POST-EDITING MODEL, AND RELATED APPARATUS
Document Type and Number:
WIPO Patent Application WO/2024/045792
Kind Code:
A1
Abstract:
The present application discloses a translation processing method, a training method for a post-editing model, and a related apparatus. The translation processing method comprises: acquiring an input sequence comprising a plurality of segments and segment identifiers, wherein the plurality of segments comprise a source language segment and a target language segment with a mask label, the segment identifiers segment the source language segment and the target language segment, and the mask label is located at a position to be recommended of the target language segment; performing embedding processing on the input sequence by means of a post-editing model, so as to obtain a word vector and a position vector corresponding to the input sequence, and obtaining a segment vector on the basis of the segment identifiers; on the basis of the word vector, the position vector and the segment vector corresponding to the input sequence, obtaining an input vector of the input sequence by means of the post-editing model; according to the input vector, performing encoding by means of the post-editing model to output an editing result; and on the basis of the editing result, performing decoding by means of the post-editing model to output a translation processing result of said position. The present application improves the accuracy of the translation processing result so as to improve the post-editing accuracy and effect.

Inventors:
YANG ZHEN (CN)
MENG FANDONG (CN)
Application Number:
PCT/CN2023/101610
Publication Date:
March 07, 2024
Filing Date:
June 21, 2023
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
TENCENT TECH SHENZHEN CO LTD (CN)
International Classes:
G06F40/58
Foreign References:
CN110765791A2020-02-07
CN113761946A2021-12-07
Other References:
YANG ZHEN, MENG FANDONG, ZHANG YINGXUE, LI ERNAN, ZHOU JIE: "WeTS: A Benchmark for Translation Suggestion", ARXIV (CORNELL UNIVERSITY), CORNELL UNIVERSITY LIBRARY, ARXIV.ORG, ITHACA, 14 March 2022 (2022-03-14), Ithaca, XP093144896, [retrieved on 20240325], DOI: 10.48550/arxiv.2110.05151
Attorney, Agent or Firm:
SHENPAT INTELLECTUAL PROPERTY AGENCY (CN)
Download PDF: