This is the official implementation for our paper Multi-Behavior Hypergraph-Enhanced Transformer for Next-Item Recommendation, accepted by KDD'22.
The code is built on Pytorch and the RecBole benchmark library. Run the following code to satisfy the requeiremnts by pip:
pip install -r requirements.txt
-
Download the three public datasets we use in the paper at: https://drive.google.com/file/d/1OFT_5Xp_az-GSHIl7QEPB9zhulbooLzE/view?usp=sharing
-
Unzip the datasets and move them to ./dataset/
-
You may also refer to the raw data here :)
python run_MBHT.py --model=[MBHT] --dataset=[tmall_beh] --gpu_id=[0] --batch_size=[2048], where [value] means the default value.
- Note that we modified the evaluation sampling setting in
recbole/sampler/sampler.pyto make it static. - The model code is at
recbole/model/sequential_recommender/mbht.py. - Feel free to explore other baseline models provided by the RecBole library and directly run them to compare the performances.
If you find our work helpful, please kindly cite our research paper:
@inproceedings{yang2022mbht,
title={Multi-behavior hypergraph-enhanced transformer for sequential recommendation},
author={Yang, Yuhao and Huang, Chao and Xia, Lianghao and Liang, Yuxuan and Yu, Yanwei and Li, Chenliang},
booktitle={Proceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining},
pages={2263--2274},
year={2022}
}