Answer Selection for Python (Sept, 2017)

Code for the paper A Hybrid Approach to Answer Selection in Question Answering Systems (IUKM, 2018)

Requirement

• 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.

Run the model

• 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

Contributors

• Phuc H. Duong - huuphucduong@gmail.com
• Hien T. Nguyen - hiennguyenthanh@gmail.com
• Duy D. Nguyen - duy.nguyen@newai.vn
• Hao T. Do - hao.do@newai.vn