Code for the paper A Hybrid Approach to Answer Selection in Question Answering Systems (IUKM, 2018)
- Source code is written in Python 3.5 (Anaconda Environment), and requires Keras (TensorFlow [GPU] back-end)
- Use the pre-trained word2vec, published by Mikolov et al.
- Download all pre-processed datasets in
data
directory - Run this command for evaluating Feed-forward neural network model:
python ann.py
- Run this command for evaluating Convolutional neural network model:
python conv1d.py
- All above programs perform three phases: (1) build the model, (2) compute
accuracy
on training and testing datasets, and (3) save model to disk
- Phuc H. Duong -
[email protected]
- Hien T. Nguyen -
[email protected]
- Duy D. Nguyen -
[email protected]
- Hao T. Do -
[email protected]