Mining translations from the web of open linked data
John Philip McCrae, Philipp Cimiano · 2013
In this paper we consider the prospect of extracting translations for words from the web of linked data. By searching for entities that have labels in both English and German we extract 665,000 translations. We then also consider a linguistic linked data resource, lemonUby, from which we extract a further 115,000 translations. We combine these translations with the Moses statistical machine translation, and we show that the translations extracted from the linked data can be used to improve the translation of unknown words.