: Vocabulaire du TAL: raw translation

applied NLP > machine translation > raw translation

raw translation

The output of a machine translation system without any post-editing or human intervention.

For example Garcia (2011) showed that even though post-editing raw translations does not always lead to significant increases in productivity this process can result in higher quality translations (when compared to translating from scratch). (Besacier & Schwartz, 2015)
Non-translators who use MT for information purposes will not even post-edit the output but use the raw translation. (Maegaard, 1997)
The main reasons to use MT are: first of all to translate documents which are less important secondly to have documents translated fast (would have preferred to send the documents to SdT but could not wait) and to use the raw translation for browsing. (Maegaard, 1997)
The natural inclination of the novice user to accept MT as a translation panacea coupled with vendor hype makes the disappointment even more acute once users see their first raw translations. (Flanagan, 1996)

http://data.loterre.fr/ark:/67375/8LP-NLWSP775-M

RDF/XML TURTLE JSON-LD Dernière modification le 29/04/2024

Loterre