recsyslearn 0.6.0

A small library to compute fairness of recommender systems.

Recsyslearn

A small library to compute fairness of recommender systems.

• Free software: GNU General Public License v3

• Documentation: https://recsyslearn.readthedocs.io.

Features

• Compute Novelty of a recommender system based on its recommendations list.

• Compute Coverage of a recommender system based on its recommendations list.

• Compute Entropy of a recommender system based on its recommendations list.

• Compute Kullback-Leibler divergence of a recommender system based on its recommendations list and the wanted target representation.

• Compute Mutual Information of a recommender system based on its recommendations list.

• Segment an implicit or explicit dataset in groups based on the activity of the users or on the popularity of the items.

Known Issues

In this version of the library, the computation of the metrics for cross groups (user and item groups together) has not been implemented yet.

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

History

0.4.0-alpha (2022-06-24)

• First release on PyPI.

0.4.1 (2022-06-27)

• Fixed novelty formula.

0.5.0 (2022-07-19)

• Added a new item segmentation method, which gives a percentage score to the items based on their popularity.

• More accurate docs, with a beautiful theme.

0.5.1 (2022-07-27)

• Fixed mantissa problem with the sum of proportion in the segmentation part.

• Improved code readability and tests coverage.

0.6.0 (2022-08-30)

• Added accuracy computation with NDCG@k