[projekt]
Klemen Arzenšek (Author), Mirjam Sepesy Maučec (Mentor), Gregor Donaj (Co-mentor)

Abstract

V projektu smo opravili pregled metrik za avtomatsko vrednotenje strojnih prevodov. Testirali smo določene metrike za ocenjevanje prevodov iz angleščine v slovenščino. Vsi testi so bili opravljeni z metrikami direktno iz Linux terminala. Testirali smo tako, da smo vsako metriko oz. sklop metrik lokalno zaganjali, vsako metriko posebej z več testnimi vzorci, da smo dobili več rezultatov in tako odpravili nepotrebna odstopanja. V poročilu bomo predstavili tudi osnove strojnega prevajanja in osnove vrednotenja le-tega. Metrike, ki smo jih uporabili pri testiranju, bomo podrobno opisali in v praksi preizkusili.

Keywords

strojno prevajanje;evalvacija;metrika avtomatske evalvacije;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: [K. Arzenšek]
UDC: 621.39
COBISS: 22109462 Link will open in a new window
Views: 897
Downloads: 75
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Analysis of metrics for automatic evaluation of machine translation
Secondary abstract: In the project, we made an overview of automatic metrics for machine translation evaluation. We tested the quality of translations from English into Slovene. All the tests were performed with metrics directly from the Linux terminal. We tested the metric locally. Each metric was tested separately with several test samples to obtain results without unnecessary deviations. The report will first present the basics of machine translation and the basics of evaluation. The metrics we have used in the testing will be described in detail and practically tested to see how they work.
Secondary keywords: machine translation;evaluation;automated evaluation metric;
URN: URN:SI:UM:
Type (COBISS): Diploma project paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Telekomunikacije
Pages: VIII, 27 f.
ID: 10960200