Uporaba evolucijskih algoritmov v statističnem in hibridnem strojnem prevajanju : doctoral dissertation

Dugonik, Jani

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Uporaba evolucijskih algoritmov v statističnem in hibridnem strojnem prevajanju : doctoral dissertation
Authors:	ID Dugonik, Jani (Author) ID Brest, Janez (Mentor) More about this mentor... ID Sepesy Maučec, Mirjam (Comentor)
Files:	DOK_Dugonik_Jani_2025.pdf (1,21 MB) MD5: BECF4CC158DF995CAE79AA3BE39A085C
Language:	Slovenian
Work type:	Doctoral dissertation
Typology:	2.08 - Doctoral Dissertation
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	Doktorska disertacija obravnava področje strojnega prevajanja visoko fleksibilnih jezikov, osredotoča pa se na izzive tako statističnega kot nevronskega strojnega prevajanja, ki jih prinašajo strukturne razlike med visoko fleksibilnimi jeziki in angleščino. Naša raziskava vključuje tudi eksperimentalni del, izveden na jezikovnem paru \mbox{slovenščina--angleščina}, ki zajema prevajanje v obe smeri. V prvem eksperimentu smo načrtovali nov pristop za optimizacijo parametrov v statističnem strojnem prevajanju z uporabo evolucijskih algoritmov. Primerjali smo sisteme statističnega strojnega prevajanja, optimizirane s klasičnimi algoritmi za optimizacijo uteži v statističnem strojnem prevajanju, in sisteme, optimizirane z evolucijskimi algoritmi. V drugem eksperimentu pa smo načrtovali in razvili hibridni pristop, ki vključuje sisteme statističnega in nevronskega strojnega prevajanja. Izvorno poved in dva ciljna prevoda, prevedena z obema sistemoma, smo pretvorili v isti vektorski prostor, iz katerega smo nato pridobili vektorje značilk. V okviru doktorske disertacije smo pred\-lagali nov nabor značilk. Z uporabo klasifikatorjev smo nato izbrali boljšega izmed dveh prevodov, statističnega in nevronskega. Evalvacijo sistemov strojnega prevajanja smo izvedli z uporabo uveljavljenih metrik, kot so BLEU, TER, chrF in COMET. Opravili smo statistično analizo eksperimentalnih rezultatov s ponovnim vzorčenjem, ki je pokazala statistično pomembne razlike v kakovosti ustvarjenih prevodov. Eksperimentalni rezultati potrjujejo, da smo s predlaganimi pristopi izboljšali kakovost strojnih prevodov.
Keywords:	evolucijski algoritem, statistično strojno prevajanje, nevronsko strojno prevajanje, hibridni pristop strojnega prevajanja, optimizacija, predstavitev besed, klasifikacija, obratno prevajanje
Place of publishing:	Maribor
Place of performance:	Maribor
Publisher:	[J. Dugonik]
Year of publishing:	2025
Number of pages:	XVI, 122 str.
PID:	20.500.12556/DKUM-88990
UDC:	004.8.021:81'322.4(043.3)
COBISS.SI-ID:	224373507
Publication date in DKUM:	29.01.2025
Views:	208
Downloads:	117
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-SA 4.0, Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-sa/4.0/
Description:	A Creative Commons license that bans commercial use and requires the user to release any modified works under this license.
Licensing start date:	04.06.2024

Secondary language

Language:	English
Title:	The use of evolutionary algorithms in statistical and hybrid machine translation
Abstract:	The doctoral dissertation tackles the field of machine translation of highly flexible languages, focusing on the challenges posed by both statistical and neural machine translation due to the structural differences between highly flexible languages and English. Our research includes an experimental part carried out on the Slovene--English language pair, covering translation in both directions. In the first experiment, we developed a new approach for parameter optimization in statistical machine translation (SMT) using evolutionary algorithms. We compared SMT systems optimized with classical weight optimization algorithms to those optimized with evolutionary algorithms. In the second experiment, we designed and developed a hybrid approach that integrates SMT and neural machine translation (NMT) systems. We converted the source sentence and the two target translations, produced by both systems, into the same vector space, from which feature vectors were then obtained. As part of the doctoral dissertation, we proposed a new set of features. Using classifiers, we selected the better translation between the SMT and NMT system outputs. The evaluation of the machine translation systems was performed using established metrics such as BLEU, TER, chrF, and COMET. We conducted a statistical analysis of the experimental results with resampling, which showed statistically significant differences in the quality of the translations created. Experimental results confirm that we have improved the quality of machine translations with the proposed approaches.
Keywords:	evolutionary algorithm, statistical machine translation, neural machine translation, hybrid machine translation approach, optimization, sentence representation, classification, backtranslation

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back