Fig 1.
NLP classification process flow.
Fig 2.
Architecture of the CBOW model.
Fig 3.
Architecture of the skip-gram model.
Fig 4.
Architecture of the LSTM network.
Fig 5.
The forget gate.
Fig 6.
The input gate.
Fig 7.
Updating the cell state.
Fig 8.
The output gate.
Fig 9.
Architecture of the BiLSTM model.
Fig 10.
Structure of the fully connected layer.
Table 1.
The experimental environment and software tools.
Fig 11.
Histogram of word counts.
Fig 12.
Data processing flow of the proposed model.
Table 2.
Table of configurable parameters.
Table 3.
Performance metrics.
Table 4.
Classification performance of the LSTM network.
Table 5.
Classification performance of the BiLSTM network.
Table 6.
Overall classification performance.
Fig 13.
Loss vs. epoch and accuracy vs. epoch on the training and validation sets.
Fig 14.
Confusion matrices of the BiLSTM and LSTM networks.
Table 7.
Classification performance of the model comparison experimental.
Fig 15.
Statistical chart of the results of the model comparison experiment.