In this deep learning-based sentence classification experiment, several methodologies were employed using different architectures to classify the sentences. The five main classifiers were constructed using the following architectures:
- A bag of words (BOW) approach with pre-trained vectors.
- A bag of words approach with randomly initialized vectors.
- A bidirectional LSTM (BiLSTM) approach with pre-trained vectors.
- A bidirectional LSTM approach with randomly initialized vectors.
- A convolutional neural network (CNN) model.
After comparing the results from each architecture, it was found that the CNN model emerged as the clear winner, demonstrating superior performance over the other models. This is likely due to the CNN's ability to effectively identify patterns and features in the input data.