MIT Press

Table 3:

Performance comparison of previous and our best models in the Yelp 2013 dataset. Our best models perform better, even though we only use a single BiLSTM encoder. Boldface correspond to the best values for each block. Underlined values correspond to the best values across the board.

Models		Acc	RMSE
UPNN (Tang et al., 2015)	CNN + word-cust + bias-cust	59.6	0.784
UPDMN (Dou, 2017)	LSTM + memory-cust	63.9	0.662
NSC (Chen et al., 2016)	LSTM + attention-cust	65.0	0.692
HCSC (Amplayo et al., 2018a)	BiLSTM + CNN + attention-cust (CSAA)	65.7	0.660
PMA (Zhu and Yang, 2017)	HierLSTM + attention-cust (PMA)	65.8	0.668
DUPMN (Long et al., 2018)	HierLSTM + memory-cust	66.2	0.667
CMA (Ma et al., 2017)	HierAttention + attention-cust (CMA)	66.4	0.677
Our best models	BiLSTM + encoder-basis-cust	66.1	0.665
	BiLSTM + bias-basis-cust	66.9	0.654
	BiLSTM + linear-basis-cust	67.1	0.662

Models		Acc	RMSE
UPNN (Tang et al., 2015)	CNN + word-cust + bias-cust	59.6	0.784
UPDMN (Dou, 2017)	LSTM + memory-cust	63.9	0.662
NSC (Chen et al., 2016)	LSTM + attention-cust	65.0	0.692
HCSC (Amplayo et al., 2018a)	BiLSTM + CNN + attention-cust (CSAA)	65.7	0.660
PMA (Zhu and Yang, 2017)	HierLSTM + attention-cust (PMA)	65.8	0.668
DUPMN (Long et al., 2018)	HierLSTM + memory-cust	66.2	0.667
CMA (Ma et al., 2017)	HierAttention + attention-cust (CMA)	66.4	0.677
Our best models	BiLSTM + encoder-basis-cust	66.1	0.665
	BiLSTM + bias-basis-cust	66.9	0.654
	BiLSTM + linear-basis-cust	67.1	0.662

View Large

This Feature Is Available To Subscribers Only