Evaluación de un sistema de traducción automática basado en reglas o por qué BLEU sólo sirve para lo que sirve

1. Mayor Martinez, Aingeru
2. Alegría Loinaz, Iñaki
3. Díaz de Ilarraza Sánchez, Arantza
4. Labaka Intxauspe, Gorka
5. Lersundi Ayestaran, Mikel
6. Sarasola Gabiola, Kepa

Export

Export

×

MLA

modern-language-association

APA

apa

Chicago

chicago-fullnote-bibliography

Harvard

harvard-cite-them-right

Vancouver

vancouver-author-date

---

RIS BibTex

Export

×

This document cannot be exported because it belongs to another portal.

Full text

lock_openRUA externo

DIALNET GOOGLE SCHOLAR lock_openRUA editor

More publications in: Procesamiento del lenguaje natural

Abstract

Matxin is a rule-based machine translation system which translates to Basque. For its evaluation we have used the HTER metric which calculates the post-editing cost, concluding that 4 of each 10 words would have to be modified to correct the output generated by the system. We have compared the quality of Matxin translations with that of a corpus based system, and the results show that Matxin performs significantly better. Given the widespread use of BLEU, we have examined the BLEU scores for both systems, and we conclude that this metric is neither effective to measure the absolute quality of a system, nor suitable to compare systems based on different strategies.

Data source: Dialnet