[Retracted] Text Sentiment Analysis of German Multilevel Features Based on Self-Attention Mechanism

Table 4

Comparison of classification results of different methods.

Model
Dostoevsky	53.40	55.45	—	—	—	—
SVM	54.23	54.80	58.45	58.62	4.22	4.01
CNN	62.51	62.64	66.54	66.65	4.03	4.01
LSTM	62.11	62.19	65.95	66.03	3.85	3.84
Bi-LSTM	62.47	62.47	66.48	66.50	4.01	4.03
BiLSTM-2layers	62.85	63.02	66.39	67.01	4.08	3.99
BiLSTM-ATT	62.59	62.57	66.56	66.63	3.97	4.06
BiLSTM-ATT2	63.4	63.36	67.85	67.96	4.45	4.6
BiLSTM-CNN	63.35	63.28	67.51	67.69	4.16	4.41
CNN-Bi-LSTM	63.47	63.51	67.83	67.98	4.36	4.47
ACBM	63.82	63.90	68.75	68.87	5.03	5.07