Table 2. Pseudocode of multi label English translation text classification algorithm.

Input: Training parameter set T = {(wk, ykj)Lj = 1}k = 1, model parameter θ.
1:	W = {wk = 1/N | k = 1, 2, …N}
2:	repeat
3:	for all w = {(Wk, ykj )Nj = 1}k = 1∈D do
4:	for l in range(N)
5:	wk = {w1, w2, …, wn}
6:	Tk = Att-BILSTM(wk)
7:	for o in rang (O)
8:	Xok = transformer(Tk)
9:	end for
10:	for p in range (p)
11:	ht1 = TLCM(Xok)
12:	ht2 = BERT(Xok)
13:	ht3 = [ht1,ht2]
14:	end for
15:	for q in range(Q)
16:	vq = Conv(ht1)
17:	Fq = Maxpooling(vq)
18:	end for
19:	M = Concat(Fq , hpt)
20:	A = Attention(M)
21:	W2 = Fully Connected (A, Dropout)
22:	yjk = δ (w2)
23:	end for
24:	Calculate the gradient of each parameter