EvalRS-CIKM-2022

Overview

This is the solution presented in the "Triplet losses-based matrix factorization for robust recommendations" paper, for the EvalRS challenge @ CIKM 2022.

While this README is currently a stub, it will contain a more detailed description of the solution later on. The work is described in more detail in the paper.

Instructions

This solution requires Python 3.x to be executed. The following are the steps to reproduce the results.

1. (optional) Create a virtual environment to install everything needed. If not created, all libraries (see next step) will be installed system-wide.

virtualenv venv
. venv/bin/activate

2. Install the dependencies in requirements.txt

pip install -r requirements.txt

Note that, although needed, PyTorch does not show up among the dependencies. This is because its installation is system-dependent (in particular for what concerns CUDA). If not already available, you may follow the instructions here.

3. Create the upload.env file, with all of the relavant information (see local.env for a list of required fields)

4. Run the code, through submission.py

python submission.py