Aspect-Level Sentiment Analysis Based on Position Features Using Multilevel Interactive Bidirectional GRU and Attention Mechanism

Table 4

Performance of different models on sentiment analysis datasets.

Dataset	Restaurant14		Laptop14		Restaurant15		Restaurant16
	Acc	F1	Acc	F1	Acc	F1	Acc	F1
LSTM	74.28	60.24	66.45	60.23	75.36	48.69	83.12	52.55
AE-LSTM	76.12	61.08	68.83	62.08	75.94	50.11	84.46	56.28
ATAE-LSTM	77.23	62.97	68.61	62.99	76.54	53.29	85.17	59.32
IAN	77.14	64.53	71.94	66.51	77.28	52.41	81.49	62.57
MemNet	80.57	65.49	71.49	68.37	77.18	53.36	81.42	60.49
PBAN	79.73	71.08	74.61	70.82	78.11	56.73	80.79	61.85
MI-biGRU	81.66	73.86	75.25	70.93	79.17	59.97	82.54	63.48

The number (9) after MemNet refers to nine computational layers that are adopted.