Table 8. Accuracies of methods (Ensemble-2) for semeval restaurant dataset.

Methods	Accuracies	Explanation
Self Attention	88.70	BERT-IL [44]
ARV, WARV and CNN	88.669 ∓ 3.60	Ensemble of ARV, WARV and CNN by Neural Network
WARV and CNN	88.630 ∓ 3.81	Ensemble of WARV and CNN by Neural Network
ARV and CNN	88.475 ∓ 1.79	Ensemble of ARV and CNN by Neural Network
XRCE	88.126	Semeval 2016 Task 5 Slot 1 Best Accuracy [1]
Hierarchical Attention	87.00	ELMo + BERT [45]
IIT-T	86.729	Semeval 2016 Task 5 Slot 1 Second Best Accuracy [1]
NileT	85.448	Semeval 2016 Task 5 Slot 1 Third Best Accuracy
IHS-R	83.935	Semeval 2016 Task 5 Slot 1 Fourth Best Accuracy [1]
ECNU	83.236	Semeval 2016 Task 5 Slot 1 Fifth Best Accuracy [1]
WARV	83.95 ∓ 0.38	Weighted Averaged Review Vectors with Neural Network
CNN	83.120 ∓ 0.44	Convolutional Neural Network with Word Vectors
INSIG	82.072	Semeval 2016 Task 5 Slot 1 Sixth Best Accuracy [1]
ARV	82.08 ∓ 0.68	Averaged Review Vectors with Neural Network
basel	76.484	Base value of Semeval 2016 Task 5 Slot 1 [1]