A Method of Chinese-Vietnamese Bilingual Corpus Construction for Machine Translation

A bilingual corpus is vital for natural language processing problems, especially in machine translation. The larger and better quality the is, higher efficiency of resulting translation is. There are two popular approaches to building a corpus. first one automatically based on resources that available internet, typically websites. second approach construct manually. Automated construction methods being used more frequently because they less expensive there growing number websites exploit. In this paper, we use automated collection website create Chinese-Vietnamese particular, collect data multilingual dictionary (https://glosbe.com). We collected from includes than 400k sentence pairs. chose 100,000 pairs experiments. From corpus, built five datasets consisting 20k, 40k, 60k, 80k, 100k pairs, respectively. addition, additional datasets, applying word segmentation sentences original datasets. experimental results showed that: (1) relatively good with highest BLEU score 19.8, although still some issues need be addressed future works; (2) is; (3) untokenized help train models tokenized