Research Article

An Efficient Sentiment Classification Method with the Help of Neighbors and a Hybrid of RNN Models

Table 6

Cross-dataset evaluation results.

GroupMethodTraining datasetTesting datasetAccuracyF1_score

OneLSTMBG (all data)BIS (train set)94.3994.70
BG (all data)BIS (test set)94.9495.35
BG (all data)BIS (all data)94.5794.90
GRUBG (all data)BIS (train set)94.6694.94
BG (all data)BIS (test set)94.7194.10
BG (all data)BIS (all data)94.6794.97
DTSCBG (all data)BIS (train set)95.9696.23
BG (all data)BIS (test set)95.2595.64
BG (all data)BIS (all data)95.9296.21

TwoLSTMBG (train set)BIS (train set)94.4094.70
BG (train set)BIS (test set)92.8993.25
BG (train set)BIS (all data)94.6994.97
GRUBG (train set)BIS (train set)93.7894.05
BG (train set)BIS (test set)94.0694.45
BG (train set)BIS (all data)93.2193.46
DTSCBG (train set)BIS (train set)95.5095.94
BG (train set)BIS (test set)96.1496.58
BG (train set)BIS (all data)95.6896.11

Bold values indicate the best result.