This is the implementation of our paper "Estimating Causal Effects on Networked Observational Data via Representation Learning", published at CIKM'22.
The original data are from this repo, kudos for the authors!
Due to the size limit, we put 1) data simulation code 2) the original data and 3) the simulated data in google drive.
We use METIS
to partion a graph. If you'd like to apply it to your data, please refer to the official package. There is also a python version.
-
Step0 (data):
mkdir data
under the root folder.- Download the simulated data and put them under the
data
folder.
-
Step1 (run):
cd ./src
- For BC dataset:
python main.py --dataset BC
- For Flickr dataset:
python main.py --dataset Flickr
- See explanations for other arguements and parameters in
main.py
.
The prediction, evluation results and embeddings are stored under the result
folder.
Song Jiang [email protected]
@inproceedings{netest2022,
title={Estimating Causal Effects on Networked Observational Data via Representation Learning},
author={Song Jiang, Yizhou Sun},
booktitle={Proceedings of the 31st ACM International Conference on Information & Knowledge Management},
year={2022}
}