Unsupervised Machine Translation: How Machines Learn to Understand Across Languages

For decades, machine translation between natural languages fundamentally relied on human-translated documents known as parallel texts, which provide direct correspondences between source and target sentences. The notion that translation systems could be trained on non-parallel texts, independently w...

Mô tả đầy đủ

Đã lưu trong:
Chi tiết về thư mục
Tác giả chính: Kvapilíková, Ivana
Định dạng: Online
Ngôn ngữ:Tiếng Anh
Được phát hành: Karolinum Press 2025
Những chủ đề:
Truy cập trực tuyến:https://directory.doabooks.org/handle/20.500.12854/161895
Các nhãn: Thêm thẻ
Không có thẻ, Là người đầu tiên thẻ bản ghi này!
_version_ 1869519335716814848
author Kvapilíková, Ivana
author_browse Kvapilíková, Ivana
author_facet Kvapilíková, Ivana
author_sort Kvapilíková, Ivana
collection Directory of Open Access Books
description For decades, machine translation between natural languages fundamentally relied on human-translated documents known as parallel texts, which provide direct correspondences between source and target sentences. The notion that translation systems could be trained on non-parallel texts, independently written in different languages, was long considered unrealistic. Fast forward to the era of large language models (LLMs), and we now know that given their sufficient computational resources, LLMs exploit incidental parallelism in their vast training data, i.e., they identify parallel messages across languages and learn to translate without explicit supervision. LLMs have since demonstrated the ability to perform translation tasks with impressive quality, rivaling systems specifically trained for translation. This monograph explores the fascinating journey that led to this point, focusing on the development of unsupervised machine translation. Long before the rise of LLMs, researchers were exploring the idea that translation could be achieved without parallel data. Their efforts centered on motivating models to discover cross-lingual correspondences through various techniques, such as the mapping of word embedding spaces, back-translation, or parallel sentence mining. Although much of the research described in this monograph predates the mainstream adoption of LLMs, the insights gained remain highly relevant. They offer a foundation for understanding how and why LLMs are able to translate.
format Online
id doab-20.500.12854ir-161895
institution Directory of Open Access Books
language eng
publishDate 2025
publishDateRange 2025
publishDateSort 2025
publisher Karolinum Press
publisherStr Karolinum Press
record_format ojs
spelling doab-20.500.12854ir-1618952025-07-02T09:28:44Z Unsupervised Machine Translation: How Machines Learn to Understand Across Languages Kvapilíková, Ivana linguistics translation language LLM machine translation thema EDItEUR::C Language and Linguistics::CF Linguistics::CFX Computational and corpus linguistics For decades, machine translation between natural languages fundamentally relied on human-translated documents known as parallel texts, which provide direct correspondences between source and target sentences. The notion that translation systems could be trained on non-parallel texts, independently written in different languages, was long considered unrealistic. Fast forward to the era of large language models (LLMs), and we now know that given their sufficient computational resources, LLMs exploit incidental parallelism in their vast training data, i.e., they identify parallel messages across languages and learn to translate without explicit supervision. LLMs have since demonstrated the ability to perform translation tasks with impressive quality, rivaling systems specifically trained for translation. This monograph explores the fascinating journey that led to this point, focusing on the development of unsupervised machine translation. Long before the rise of LLMs, researchers were exploring the idea that translation could be achieved without parallel data. Their efforts centered on motivating models to discover cross-lingual correspondences through various techniques, such as the mapping of word embedding spaces, back-translation, or parallel sentence mining. Although much of the research described in this monograph predates the mainstream adoption of LLMs, the insights gained remain highly relevant. They offer a foundation for understanding how and why LLMs are able to translate. Published 2025-07-02T09:28:42Z 2025-07-02T09:28:42Z 2025-06 chapter 9788024660783 https://directory.doabooks.org/handle/20.500.12854/161895 eng image/jpeg Attribution 4.0 International https://karolinum.cz/knihy/kvapilikova-unsupervised-machine-translation-how-machines-learn-to-understand-across-languages-31561 https://dspace.cuni.cz/bitstream/handle/20.500.11956/198746/9788024660844.pdf Karolinum Press 10.14712/9788024660844 10.14712/9788024660844 8d94c6aa-2a71-4559-9698-bf3e883a574a 9788024660783 176 Prague open access
spellingShingle linguistics
translation
language
LLM
machine translation
thema EDItEUR::C Language and Linguistics::CF Linguistics::CFX Computational and corpus linguistics
Kvapilíková, Ivana
Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title_full Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title_fullStr Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title_full_unstemmed Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title_short Unsupervised Machine Translation: How Machines Learn to Understand Across Languages
title_sort unsupervised machine translation how machines learn to understand across languages
topic linguistics
translation
language
LLM
machine translation
thema EDItEUR::C Language and Linguistics::CF Linguistics::CFX Computational and corpus linguistics
topic_facet linguistics
translation
language
LLM
machine translation
thema EDItEUR::C Language and Linguistics::CF Linguistics::CFX Computational and corpus linguistics
url https://directory.doabooks.org/handle/20.500.12854/161895
work_keys_str_mv AT kvapilikovaivana unsupervisedmachinetranslationhowmachineslearntounderstandacrosslanguages