This repository accompanies the real-world experiments conducted in the paper "Unbiased Pairwise Learning from Biased Implicit Feedback" by Yuta Saito, which has been accepted by ICTIR'20.
- python>=3.7
- numpy==1.18.1
- pandas==0.25.1
- scikit-learn==0.23.1
- tensorflow==1.15.2
- pyyaml==5.1.2
To run the simulation with real-world datasets, the following datasets need to be prepared as described below.
- download the Yahoo! R3 dataset and put
train.txtandtest.txtfiles into./data/yahoo/raw/directory. - download the Coat dataset and put
train.asciiandtest.asciifiles into./data/coat/raw/directory.
First, to preprocess the datasets, navigate to the src/ directory and run the command
python preprocess_datasets.py -d coat yahooThen, run the following command in the same directory
for data in yahoo coat
do
for model in wmf relmf bpr ubpr upl_bpr relmf_du ubpr_nclip
do
python main.py -m $model -d $data -r 10
done
doneThis will run real-world experiments conducted in Section 4.
After running the experimens, you can summarize the results by running the following command in the src/ directory.
python summarize_results.py -d yahoo coatOnce the code is finished executing, you can find the summarized results in ./paper_results/ directory.
We thank Minato Sato for his helpful comments, discussions, and advice.