Welcome to TDM!

This is an official implementation of Addressing Missing Data Issue for Diffusion-based Recommendation, which is accepted by SIGIR'2025.

Algorithms

The training and generating phases of TDM are as follows.

Reproduce the results

Our experiment settings are specified in the TDM.yml file. To create the environment, you can use the following command:

conda env create -f TDM.yml

Zhihu

nohup python -u TDM.py --data zhihu --timesteps 1000 --lr 0.01 --beta_sche linear --w 6 --cuda 6 --eval 5 --optimizer adamw --diffuser_type mlp1 --random_seed 100 >> log/TDM_zhihu.log 2>&1 &

YooChoose

nohup python -u TDM.py --data yc --timesteps 2000 --lr 0.0001 --beta_sche linear --w 0 --cuda 7 --optimizer adamw --diffuser_type mlp1 --random_seed 100 >> log/TDM_yc.log 2>&1 &

KuaiRec

nohup python -u TDM.py --data ks --eval 5 --epoch 30 --timesteps 2000 --lr 0.00005 --beta_sche linear --w 2 --cuda 4 --optimizer adamw --diffuser_type mlp1 --random_seed 100 --linespace 100 >> log/TDM_ks.log 2>&1 &

Citation

❤ If you find our repository useful in your research, please star us ⭐ and consider citing:

@inproceedings{mao2025TDM,
  title={Addressing Missing Data Issue for Diffusion-based Recommendation},
  author={Wenyu Mao, Zhengyi Yang, Jiancan Wu, Haozhe Liu, Yancheng Yuan, Xiang Wang, Xiangnan He},
  booktitle={Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2025},
  year={2024}
}