This is a Pytorch implementation of the model described in our paper:
Z. Wang, Q. Xu, Z. Yang, X. Cao and Q. Huang. Implicit Feedbacks are Not Always Favorable: Iterative Relabeled One-Class Collaborative Filtering against Noisy Interactions. MM2021.
- Pytorch >= 1.5.1
- numpy
We convert the datasets ML100K
, ML1M
and Netflix
to our train and test files in the data/
folder.
To generate implicit training data, we randomly select
The ratings are stored in the files *.lsvm. The data format of the line user_id is item_id:rating. The numeric ratings range from 1 to 5.
Here is an example to generate the new data and train the model.
python main.py
Please cite our paper if you use this code in your own work:
@inproceedings{DBLP:conf/mm/WangX0CH21,
author = {Zitai Wang and
Qianqian Xu and
Zhiyong Yang and
Xiaochun Cao and
Qingming Huang},
title = {Implicit Feedbacks are Not Always Favorable: Iterative Relabeled One-Class Collaborative Filtering against Noisy Interactions},
booktitle = {{ACM} Multimedia Conference},
pages = {3070--3078},
year = {2021},
}