MIT Press

Skip to Main Content

Table 2

Comparisons with different baselines for single-domain aspect/opinion terms extraction.

Models	R		L		D
Models	AS	OP	AS	OP	AS	OP
CRF-1	67.50	73.49	67.38	72.07	36.72	55.44
CRF-2	70.10	73.76	68.30	71.36	39.18	57.67
LSTM	74.48	77.05	72.48	73.05	50.52	63.14

CRF-embedding	72.10	75.94	70.22	73.56	46.78	53.44
SRNN	68.15	68.08	64.10	62.37	38.53	50.83
RNCRF	76.78	79.20	72.28	73.74	49.40	63.77

SRNN-GRU	75.70	79.34	73.62	73.31	50.28	64.15

Models	R		L		D
Models	AS	OP	AS	OP	AS	OP
CRF-1	67.50	73.49	67.38	72.07	36.72	55.44
CRF-2	70.10	73.76	68.30	71.36	39.18	57.67
LSTM	74.48	77.05	72.48	73.05	50.52	63.14

CRF-embedding	72.10	75.94	70.22	73.56	46.78	53.44
SRNN	68.15	68.08	64.10	62.37	38.53	50.83
RNCRF	76.78	79.20	72.28	73.74	49.40	63.77

SRNN-GRU	75.70	79.34	73.62	73.31	50.28	64.15