UNIGE document Master
previous document  unige:19363  next document
add to browser collection

Statistical Machine Translation and Controlled Language Impact and Interactions

Denomination Master universitaire en traduction mention technologies de la traduction
Defense Maîtrise : Univ. Genève, 2012
Abstract This thesis explores the possible interactions between controlled language and statistical machine translation. Building on some previous research done at Autodesk Development Sàrl., we conducted a separate short research to determine whether it would be possible to improve SMT output by editing the input texts according to CL rules (in particular, AcrocheckTM rules). Although these results are not conclusive due to the limited extent of the tests performed, this preliminary study has shown that the impact of CL on SMT systems (particularly, Moses) is slightly positive, rarely negative. The perspectives are very optimistic and research in this sense may bear fruits. Future lines of research are suggested.
Full text
Master (2.3 MB) - public document Free access
(ISO format)
PORRO RODRIGUEZ, Victoria. Statistical Machine Translation and Controlled Language Impact and Interactions. Université de Genève. Maîtrise, 2012. https://archive-ouverte.unige.ch/unige:19363

444 hits



Deposited on : 2012-04-03

Export document
Format :
Citation style :