In this paper, we presented the process of building a good quality bilingual corpus for a low-resource language pair and proposed a novel method of sentence alignment that takes advantage of pre-trained modern models for rich-resource languages. In our experiments on aligning sentences and building a bilingual corpus for the Vietnamese-Laos language pair, we achieved higher precision and recall than other good sentence alignment methods and a good quality sentence-aligned Vietnamese-Laos bilingual corpus. |
