Graph2Seq-for-KGQG

Code & data accompanying the paper "Toward Subgraph Guided Knowledge Graph Question Generation with Graph Neural Networks".

Architecture

Get started

Prerequisites

This code is written in python 3. You will need to install a few python packages in order to run the code. We recommend you to use virtualenv to manage your python packages and environments. Please take the following steps to create a python virtual environment.

• If you have not installed virtualenv, install it with pip install virtualenv.
• Create a virtual environment with virtualenv venv.
• Activate the virtual environment with source venv/bin/activate.
• Install the package requirements with pip install -r requirements.txt.

In order to compute the meteor score, please download the required data from here and put it under the src/core/evaluation/meteor/data folder.

Run the QG model

• Download the pretrained GloVe word ebeddings glove.840B.300d.zip and move glove.840B.300d.txt to the data folder in this repo.
• Download the data from here and move it to the data folder in this repo.
• Cd into the src folder
• Run the QG model and report the performance

      python main.py -config config/mhqg-wq/graph2seq.yml
      python main.py -config config/mhqg-pq/graph2seq.yml
  

• You can finetune the above trained QG model using RL by running the following command:

      python main.py -config config/mhqg-wq/rl_graph2seq.yml
      python main.py -config config/mhqg-pq/rl_graph2seq.yml
  

• You can find the output data in the out_dir folder specified in the config file.

Reference

If you found this code useful, please consider citing the following paper:

Chen, Yu, Lingfei Wu, and Mohammed J. Zaki. "Toward Subgraph Guided Knowledge Graph Question Generation with Graph Neural Networks." arXiv preprint arXiv:2004.06015 (2020).

@article{chen2020toward,
  title={Toward Subgraph Guided Knowledge Graph Question Generation with Graph Neural Networks},
  author={Chen, Yu and Wu, Lingfei and Zaki, Mohammed J.},
  journal={arXiv preprint arXiv:2004.06015},
  year={2020}
}