This thesis is concerned with the problem of translating Italian connectives into LIS, the Italian Sign Language. We explain the interest and the unique challenges of this problem, and we introduce the ATLAS project for automatic translation from Italian to LIS, which forms the backdrop of our work. We then detail the analysis we performed on the small bilingual corpus of Italian and LIS sentences that was provided by the ATLAS project. We propose an alignment method between the Italian and LIS syntax trees that helps identify the effect that an Italian connective has on the LIS translation of the sentence. This results in four possible ways that a connective can be translated: with a corresponding sign, by affecting the location or shape of other signs, by affecting the LIS syntax tree, or by being omitted altogether. A clustering process applied to these result groups produces a categorization of connectives that is linguistically meaningful. By training a decision tree based classifier, we are able to extract rules to determine how to translate a given connective, with the aim of integrating these rules into the ATLAS translation pipeline. Finally, we evaluate the performance of our approach in comparison with the ATLAS pipeline.
Translating Italian connectives into Italian sign language
LUGARESI, CAMILLO
2012/2013
Abstract
This thesis is concerned with the problem of translating Italian connectives into LIS, the Italian Sign Language. We explain the interest and the unique challenges of this problem, and we introduce the ATLAS project for automatic translation from Italian to LIS, which forms the backdrop of our work. We then detail the analysis we performed on the small bilingual corpus of Italian and LIS sentences that was provided by the ATLAS project. We propose an alignment method between the Italian and LIS syntax trees that helps identify the effect that an Italian connective has on the LIS translation of the sentence. This results in four possible ways that a connective can be translated: with a corresponding sign, by affecting the location or shape of other signs, by affecting the LIS syntax tree, or by being omitted altogether. A clustering process applied to these result groups produces a categorization of connectives that is linguistically meaningful. By training a decision tree based classifier, we are able to extract rules to determine how to translate a given connective, with the aim of integrating these rules into the ATLAS translation pipeline. Finally, we evaluate the performance of our approach in comparison with the ATLAS pipeline.File | Dimensione | Formato | |
---|---|---|---|
2013_07_Lugaresi.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Thesis text
Dimensione
1.22 MB
Formato
Adobe PDF
|
1.22 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/80833