- Usage of bi-directional GRU networks.
- Mechanism for re-sampling of data, as their is a huge imbalance in the number of classes.
- Usage of CapsNet.
- Usage of different embeddings, and how they can be combined together to provide better results.
- Download the dataset from kaggle and place them in the same directory as the notebook.
- Specify the path of embeddings in the EMBEDDINGS_FILES list.
- Keras/ Tensorflow
- sklearn
- Pandas
- Numpy