Python recommendation tools

LensKit is a set of Python tools for experimenting with and studying recommender systems. It provides support for training, running, and evaluating recommender algorithms in a flexible fashion suitable for research and education.

LensKit for Python (LKPY) is the successor to the Java-based LensKit project.

[!IMPORTANT] If you use LensKit for Python in published research, please cite:

Michael D. Ekstrand. 2020. LensKit for Python: Next-Generation Software for Recommender Systems Experiments. In Proceedings of the 29th ACM International Conference on Information and Knowledge Management (CIKM '20). DOI:10.1145/3340531.3412778. arXiv:1809.03125 [cs.IR].

[!WARNING] This is the main branch of LensKit, following new development in preparation for the 2024 release. It will be changing frequently and incompatibly. You probably want to use a stable release

Installing

To install the current release with Anaconda (recommended):

conda install -c conda-forge lenskit

Or you can use pip:

pip install lenskit

To use the latest development version, install directly from GitHub:

pip install -U git+https://github.com/lenskit/lkpy

Then see Getting Started

Developing

To contribute to LensKit, clone or fork the repository, get to work, and submit a pull request. We welcome contributions from anyone; if you are looking for a place to get started, see the [issue tracker][].

Our development workflow is documented in the wiki; the wiki also contains other information on developing LensKit. User-facing documentation is at https://lkpy.lenskit.org.

We recommend using an Anaconda environment for developing LensKit. We don't maintain the Conda environment specification directly - instead, we maintain information in pyproject.toml to be able to generate it, so that we define dependencies and versions in one place.

conda-lock can help you set up the environment; the LensKit build tools automate this.

# install bootstrap enviroinment
conda env create -n lkboot -f https://raw.githubusercontent.com/lenskit/lkbuild/main/boot-env.yml
# create the lock file for Python 3.10
conda run -n lkboot --no-capture lkbuild dev-lock -v 3.10
# create the environment
conda env create -n lkpy -f conda-linux-64.lock

This will create a Conda environment called lkpy with the packages required to develop and test LensKit.

Testing Changes

You should always test your changes by running the LensKit test suite:

python -m pytest

If you want to use your changes in a LensKit experiment, you can locally install your modified LensKit into your experiment's environment. We recommend using separate environments for LensKit development and for each experiment; you will need to install the modified LensKit into your experiment's repository:

conda activate my-exp
conda install -c conda-forge flit
cd /path/to/lkpy
flit install --pth-file --deps none

You may need to first uninstall LensKit from your experiment repo; make sure that LensKit's dependencies are all still installed.

Once you have pushed your code to a GitHub branch, you can use a Git repository as a Pip dependency in an environment.yml for your experiment, to keep using the correct modified version of LensKit until your changes make it in to a release.

Resources

• Documentation
• Mailing list, etc.

Acknowledgements

This material is based upon work supported by the National Science Foundation under Grant No. IIS 17-51278. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.