Bilingual seed lexicon adaptation for entity translation extraction

Bilingual seed lexicon, which is considered as a bridge between two languages, is one of the main resources used for entity translation extraction tasks from comparable corpora. However, little attention has been paid to this lexicon except its coverage. In fact, the quality of the seed lexicon is one of the key factors that affect the accuracy of entity translation extraction. In this paper, we propose a new self-adaptive model. We use a word segmentation technique to adapt segmented corpora and then propose two strategies of weight allocation and corresponding filter. Experiments demonstrate that our technique significantly outperforms the standard approach.