Creating high-quality , large-scale bilingual knowledge bases using minimal resources

We describe a complete production cycle for the semi-automatic development of bilingual knowledge bases. The methodology is based on the optional availability of a source language monolingual corpus and on employing lexicographers who only need to be bilingual and computer literate. Bilingual resources (corpora, Machine Readable Dictionaries, etc.) are not required, and the employment of skilled and trained personnel is minimized, thus making the approach suitable for minority languages. The methodology is language independent and has been extensively used in generating an English-Spanish bilingual lexicon for Machine Translation.