Multi-Self-Attention for Aspect Category Detection and Biomedical Multilabel Text Classification with BERT

Table 4

The test set results for the Chinese EMRC task.

Model	Precision	Recall	F1
ML-KNN	71.85	84.09	76.94
Rank SVM	81.62	84.83	82.68
TextCNN	95.20	92.60	93.00
BiLSTM-Attention	92.55	91.95	89.22
Attention-XML	92.64	92, 55	89.78
BERT base	95.55	93.33	93.41
BERT Att	94.97	95.28	93.48
BERT-MSA	96.75	94.57	93.94