Automating the construction of bilingual terminology lexicons

Translation is a labor intensive process. We propose a general methodology for automatic terminology extraction and alignment which could substantially reduce the translator's workload. The goal is to take advantage of existing technology in terminology extraction and statistical word alignment to automatically construct a bilingual terminology lexicon by exploiting bilingual parallel aligned corpora. This paper introduces the technology in each area and discusses some simple heuristic methods for using the output from each component to build a bilingual terminology lexicon. The process is illustrated by an in-depth analysis of a single sentence pair.