Scholarlink

This paper describes the methodology of development of a hybrid machine translation (MT) model for bidirectional Sindhi–English neural MT (NMT) systems. We trained the system using a transfer-based approach which then used NMT model training. We used the Sindhi–English language pair for this study and trained the model for Sindhi–English as well as English–Sindhi language pairs. Models thus developed were evaluated using standard MT evaluation which produced good results as compared to human judgements.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploring the Feasibility of a Transfer-Based Approach in the Era of Deep Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nisheeth Joshi,
  • Palak Arora,
  • Bharti Nathani,
  • Chandra Prakash Dadlani

摘要

This paper describes the methodology of development of a hybrid machine translation (MT) model for bidirectional Sindhi–English neural MT (NMT) systems. We trained the system using a transfer-based approach which then used NMT model training. We used the Sindhi–English language pair for this study and trained the model for Sindhi–English as well as English–Sindhi language pairs. Models thus developed were evaluated using standard MT evaluation which produced good results as compared to human judgements.