Neural machine translation for Indian language pair using hybrid attention mechanism

Machine translation is an effective way of achieving suitable translations for a given language pair without human intervention. Machine translation can be a very tedious task when the resources for the concerned language pair are low, and the English–Assamese language pair is one such. Neural machi...

Full description

Saved in:
Bibliographic Details
Published in:Innovations in systems and software engineering 2024-06, Vol.20 (2), p.175-183
Main Authors: Nath, Basab, Sarkar, Sunita, Das, Surajeet, Mukhopadhyay, Somnath
Format: Article
Language:English
Subjects:
Artificial Intelligence
Artificial neural networks
Computer Applications
Computer Science
English language
Language
Machine translation
S.I. : Multifaceted Intelligent Computing Systems (MICS)
Software Engineering
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Machine translation is an effective way of achieving suitable translations for a given language pair without human intervention. Machine translation can be a very tedious task when the resources for the concerned language pair are low, and the English–Assamese language pair is one such. Neural machine translation is among the different machine translation techniques that use artificial neural networks to learn the nonlinear relationship between bilingual sentence pairs. In this paper, a neural machine translation model is implemented, which uses our proposed hybrid attention mechanism. This hybrid attention mechanism consists of additive and multiplicative attention mechanisms. The proposed attention mechanism is not only capable of selecting an optimal attention mechanism, but also gives the average loss value per epoch to be minimal. The model with our proposed hybrid attention mechanism outperforms the other two models with additive and multiplicative attention by achieving a BLEU score of 27.20 on English–Assamese and a BLEU score of 25.88 on the Assamese–English language pair.
ISSN:1614-5046
1614-5054
DOI:10.1007/s11334-021-00429-z