Naslov
Idioms in state-of-the-art Croatian-English and English-Croatian SMT systems

Autori
Manojlović, Maja ; Dajak, Luka ; Brkić Bakarić, Marija

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the MIPRO 2017 40th Jubilee International Convention / Petar Biljanović - Rijeka, 2017, 1798-1802

ISBN
978-953-233-093-9

Skup
Proceedings of the MIPRO 2017 40th Jubilee International Convention

Mjesto i datum
Opatija, Hrvatska, 22.05.2017. - 26.05.2017

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
idioms, Croatian-English, SMT, English-Croatian, machine translation

Sažetak
Idioms are well known for posing problems to non-native speakers, let alone machines. A failure to identify idioms often leads to unnatural, even hilarious outputs. This paper investigates the treatment of idioms in state- of-the art SMT systems involving English and Croatian. First we introduce the concept of idioms. Then we construct three short stories abundant with idioms per each language, and translate them into the other language by two state-of-the-art SMT systems. Next we manually inspect the outputs and present results. For the purpose of conducting analysis, we devise an error taxonomy for handling idioms.

Izvorni jezik
Engleski

Znanstvena područja
Informacijske i komunikacijske znanosti

POVEZANOST RADA