This is our PyTorch implementation for the paper:
Jingxuan Wen, Huafeng Liu and Liping Jing. 2023. Modeling Preference as Weighted Distribution over Functions for User Cold-start Recommendation. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management.
- To characterize the uncertainty in the user decision process, we model user preference as weighted distribution over functions, with the aid of neural processes.
- To capture the global intent and obtain a more stable learning process, we further consider intra-user uncertainty and inter-user importance respectively.
- We provide a theoretical explanation that why the proposed model performs well than regular neural process based recommendation methods.
- Extensive experiments have been conducted on four wildly used benchmark datasets, demonstrating significant improvements over several state-of-the-art baselines.
The code has been tested running under Python 3.7.10. The required packages are as follows:
- pytorch == 1.4.0
- numpy == 1.20.2
- scipy == 1.6.3
- tqdm == 4.60.0
- bottleneck == 1.3.4
- pandas ==1.3.4
The parameters have been clearly introduced in main.py
.
-
Last.FM dataset
python main.py --dataset=lastfm --gpu_id=0 --l_max=50
-
ML 1M dataset
python main.py --dataset=ml1m --lr=1e-4 --n_epoch=100
-
Epinions dataset
python main.py --dataset=epinions --n_epoch=15
-
Yelp dataset
python main.py --dataset=yelp --n_epoch=10