Evaluating Machine Translation of Latin Interjections in the Digital Library of Polish and Poland-related News Pamphlets

被引：2

作者：

Ogrodniczuk, Maciej ^{[1
]}

Krynska, Katarzyna ^{[2
]}

机构：

[1] Polish Acad Sci, Inst Comp Sci, Jana Kazimierza 5, PL-01248 Warsaw, Poland

[2] Polish Acad Sci, Inst Polish Language, Al Mickiewicza 31, PL-31120 Krakow, Poland

来源：

FROM BORN-PHYSICAL TO BORN-VIRTUAL: AUGMENTING INTELLIGENCE IN DIGITAL LIBRARIES, ICADL 2022 | 2022年 / 13636卷

关键词：

Digital library; Machine translation; Evaluation; Middle Polish;

D O I：

10.1007/978-3-031-21756-2_34

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

In The Digital Library of Polish and Poland-related Ephemeral Prints from the 16th, 17th and 18th Centuries a small fraction of items contains manually created Latin-Polish dictionaries explaining Latin fragments injected into Polish content. At the same time, rapid development of machine translation creates new opportunities for creating such dictionaries automatically. In this paper, we verify whether existing translation solutions are already capable of generating useful results in this Latin-Polish setting. We investigate two systems available for this language pair: the familiar Google neural engine and the GPT3 model, then we test the translation of isolated and context-embedded phrases and evaluate its results with both automatic and human metrics: BLEU and White's 5-point scale of adequacy and fluency.

引用

页码：430 / 439

页数：10