DAN

Code for paper:

Bei Wang, Chenrui Zhang, Hao Zhang, Xiaoqing Lyu, Zhi Tang, Dual Autoencoder Network with Swap Reconstruction for Cold-Start Recommendation (CIKM2020)

In this paper, we propose an end-to-end Dual Autoencoder Network (DAN) for user cold-start recommendations with a pair of encoder-decoder networks. Conceptually, the proposed encoder in each domain adopts a graph neural network to embed the high-order collaborative information among users and items in the interaction graph via multi-hop propagation for effective user preference learning. The decoder transforms the user information to the other domain for recommendations.

Requirement

• PyTorch
• dgl
• tensorboardX

Dataset

The data set used in this paper is the public data set, which can be found on the website.

How to run

cd src
sh run.sh

How to cite

@inproceedings{wang2020dual,
  title={Dual autoencoder network with swap reconstruction for cold-start recommendation},
  author={Wang, Bei and Zhang, Chenrui and Zhang, Hao and Lyu, Xiaoqing and Tang, Zhi},
  booktitle={Proceedings of the 29th ACM International Conference on Information \& Knowledge Management},
  pages={2249--2252},
  year={2020}
}