Knjižnica Filozofskog fakulteta
Sveučilišta u Zagrebu
Faculty of Humanities and Social Sciences Institutional Repository

Analiza sustava za evaluaciju strojnih prijevoda primjenom Dinamičkog okvira i Višedimenzionalne metrike

Downloads

Downloads per month over past year

Kranjčić, Denis. (2016). Analiza sustava za evaluaciju strojnih prijevoda primjenom Dinamičkog okvira i Višedimenzionalne metrike. Diploma Thesis. Filozofski fakultet u Zagrebu, Department of Information Science. [mentor Seljan, Sanja].

[img]
Preview
PDF (Croatian)
Download (2MB) | Preview

Abstract

This master's thesis deals with an analysis of systems for machine translation evaluation and consists of a theoretical and a practical section. The paper starts with a general description of the applications of machine translation and a few practical examples. This is followed by a presentation of different types of machine translation evaluation, which are most broadly divided into automatic and human methods of evaluation. After that, the paper focuses on two platforms which provide tools for human evaluation of MT: TAUS Dynamic Quality Framework and Multidimensional Quality Metrics. These tools are described in great detail, along with the process of setting up and performing evaluation tasks. For the practical part, an experimental research has been conducted in which most of the described tools were used. The research consists of an evaluation of machine translations of texts from two different domains performed by online tools Google Translate and Bing Translator. In the concluding section, the research results are analysed, as well as the functionalities of the tools used. At the end of the paper there is a list of works cited, a list of pictures, tables and graphs, and an appendix.

Item Type: Diploma Thesis
Uncontrolled Keywords: machine translation, machine translation evaluation, MQM, DQF, human evaluation, Croatian-English language pair
Subjects: Information sciences > Social-humanistic informatics
Departments: Department of Information Science
Supervisor: Seljan, Sanja
Date Deposited: 22 Feb 2017 08:59
Last Modified: 22 Feb 2017 08:59
URI: http://darhiv.ffzg.unizg.hr/id/eprint/7955

Actions (login required)

View Item View Item