Statistical machine translation of croatian weather forecasts: how much data do we need? / Ljubešić, Nikola ; Bago, Petra ; Boras, Damir.
By: Ljubešić, Nikola, informatičar
.
Contributor(s): Boras, Damir [aut] | Bago, Petra [aut].
Material type: 
ArticleDescription: 303-308.ISSN: 1330-1136.Other title: Statistical Machine Translation of Croatian Weather Forecasts: How Much Data Do We Need? [Naslov na engleskom:].Subject(s): 5.04 | statistical machine translation, weather forecast, automatic evaluation, human evaluation hrv | statistical machine translation, weather forecast, automatic evaluation, human evaluation engOnline resources: Click here to access online
In:
CIT. Journal of computing and information technology 18 (2010), 4 ; str. 303-308Summary: This research is the first step towards developing a system for translating Croatian weather forecasts into multiple languages. This step deals with the Croatian-English language pair. The parallel corpus consists of a one-year sample of the weather forecasts for the Adriatic, con- sisting of 7, 893 sentence pairs. Evaluation is performed by the automatic evaluation measures BLUE, NIST and METEOR, as well as by manually evaluating a sample of 200 translations. We have shown that with a small- sized training set and the state-of-the art Moses system, decod- ing can be done with 96% accuracy concerning adequacy and fluency. Additional improvement is expected by increasing the training set size. Finally, the correlation of the recorded evaluation measures is explored.
This research is the first step towards developing a system for translating Croatian weather forecasts into multiple languages. This step deals with the Croatian-English language pair. The parallel corpus consists of a one-year sample of the weather forecasts for the Adriatic, con- sisting of 7, 893 sentence pairs. Evaluation is performed by the automatic evaluation measures BLUE, NIST and METEOR, as well as by manually evaluating a sample of 200 translations. We have shown that with a small- sized training set and the state-of-the art Moses system, decod- ing can be done with 96% accuracy concerning adequacy and fluency. Additional improvement is expected by increasing the training set size. Finally, the correlation of the recorded evaluation measures is explored.
Projekt MZOS 130-1301679-1380
ENG
There are no comments for this item.