Leveraging Contextual Sentences for Text Classification by Using a Neural Attention Model

Table 6

Results of our models and other methods from the literature: HMM [16]; CNN-FF and LSTM-FF [36]; CA-LSTM: contextual attentive LSTM [40]; CR-attention: CNN + RNN + attention [38].

Models	Accuracy	Time
Our best model based on CNN	73.0/83.4	412/335
Our best model based on LSTM	73.9/85.9	1141/924
CR-attention	73.8/84.3	1320/1182
CNN + FF	73.1/84.6	732/620
LSTM + FF	69.6/84.3	940/762
HMM	71.0/—	210/—
CA-LSTM	72.6/—	1026/—