A Character Level Based and Word Level Based Approach for Chinese-Vietnamese Machine Translation

Table 5

Distribution of number of words and number of sentences in experimental corpora of 33,372 sentence pairs.

Corpora		NS	CL		WL
			NW	NW/NS	NW	NW/NS
Chinese	Training	30,036	301,630	10.0	221,419	7.4
	Developing	1,668	16,973	10.2	12,468	7.5
	Testing	1,668	17,049	10.2	12,453	7.5
Vietnamese	Training	30,036	316,453	10.5	278,232	9.3
	Developing	1,668	17,839	10.7	15,679	9.4
	Testing	1,668	17,745	10.6	15,617	9.4

NS is “number of sentences”, NW is “number of words”, and NW/NS is “NW per NS”.
We used BLEU score and TER score to evaluate the performance of the translation systems.