Automatic Translation of Norwegian Noun Compounds
Lars Bungum, Stephan Oepen · 2009
This paper discusses the automated translation of Norwegian nominal compounds into English, com-bining (a) compound segmentation, (b) component translation, (c) bi-lingual translation templates, and (d) probabilistic ranking. In this approach, a Nor-wegian compound will typically give rise to a large number of possible translations, and the selection of the ‘right ’ candidate is approaches as an interesting machine learning problem. Our work extends the seminal approach of Tanaka and Baldwin in several ways, including a clarification of some fine points of their earlier work, adaptation to a more adequate machine learning framework, application to a Ger-manic language with a small speech community and