A Novel Deep Learning Method for Obtaining Bilingual Corpus from Multilingual Website

Table 2

The size and accuracy of obtaining parallel sentences in different number of training corpus.

Model		The number of training parallel sentences
		2,000	5,000	10,000	20,000	40,000
LSTM	size	13,000	33,000	65,000	92,000	126,000
	accuracy	0.6	0.71	0.78	0.81	0.82
C-BiRNN	size	14,000	28,000	58,000	86,000	121,000
	accuracy	0.58	0.63	0.68	0.70	0.72