Hotel Review Classification Based on the Text Pretraining Heterogeneous Graph Neural Network Model

Table 2

Comparison of results.

Dataset	Comparative model	Accuracy	Recall	F1 score
Hotel review (our dataset)	BERT + HGAN	0.70	0.72	0.72
	BERT	0.62	0.62	0.61
	HGAN	0.51	0.46	0.59
	ERNIE2.0	0.69	0.67	0.68
	TF-IDF + SVM	0.36	0.30	0.33
	TextCNN	0.50	0.50	0.48
THUCNews	BERT + HGAN	0.83	0.84	0.84
	BERT	0.72	0.74	0.68
	HGAN	0.71	0.75	0.73
	ERNIE2.0	0.78	0.74	0.74
	TF-IDF + SVM	0.48	0.66	0.45
	TextCNN	0.57	0.83	0.68
Headlines Today News	BERT + HGAN	0.78	0.76	0.80
	BERT	0.65	0.83	0.62
	HGAN	0.72	0.80	0.68
	ERNIE2.0	0.76	0.88	0.77
	TF-IDF + SVM	0.58	0.67	0.62
	TextCNN	0.65	0.72	0.73