|Reranking Candidate Lists for Improved Lexical Induction
|Year of Publication
|Jakubina, L., and P. Langlais
|Advances in Artificial Intelligence
|Springer International Publishing
|Identifying translations in bilingual material–-also referred to as the Bilingual Lexicon Induction (BLI) task–-is a challenge that has attracted many researchers since a long time. In this paper, we investigate the reranking of two types of state-of-the-art approaches that have been used for the task. We test our reranker on four language pairs (translating from English), analyzing the influence of the frequency of the source terms we seek to translate. Our reranking approach almost invariably leads to performance gains for all the translation directions we consider.