FairBook

The recent study of Abdollahpouri et al. [1] in the domain of movie recommendation shows that the popularity bias causes unfair recommendations in respect to both the providers of less popular items and users with little interest in popular items. In this paper, we extend and reproduce the study of Abdollahpouri et al. [1] in the domain of book recommendation.

Dataset

In this study, we use Book-Crossing dataset to evalue the performance and fairness of the models. The dataset and user groups are accessible in the dataset folder. The foder contains the following files:

• user-group folder: active_users.txt, inactive_users.txt, medium_users.txt which are related to Niche, Diverse, and BestSeller user groups, respectively.
• BX-Book-Ratings.csv: The original Book-Crossing dataset which contains User-ID, ISBN, and Book-Rating.
• BX-Book-Explicit-5Rate-Map.csv: The explicit 5-core book-crossing dataset in which the uesr and item IDs are mapped into new IDs. (CSV format).
• BX-Book-Explicit-5Rate-Map.txt: The explicit 5-core book-crossing dataset in which the uesr and item IDs are mapped into new IDs. (TXT format).

BookCrossing

The statistics of the datasets after preprocessing are as follows:

• Number of users: 6,358
• Number of items: 6,921
• Number of interactions: 88,552
• User per groups: {Niche: 1271, Diverse: 3816, BestSeller: 1271},
• User Profile size per group: {Niche: 14006, Diverse: 61889, BestSeller: 12657}

Codes

We provide the FairBook.ipynb to run the code and the model for the analysis and recommendation performacne.

Reference

Please cite our paper if you use our datasets or implementations:

@inproceedings{naghiaei2022fairbook,
title = {The Unfairness of Popularity Bias in Book Recommendation},
author={Naghiaei, Mohammadmehdi and Rahmani, Hossein A. and Dehghan, Mahdi},
booktitle = {Third International Workshop on Algorithmic Bias in Search and Recommendation (Bias@ECIR 2022)},
year = {2022},
url = {https://arxiv.org/abs/XXX},
}

Team

Mohammadmehdi Naghiaei, University of Southern California

Hossein A. Rahmani, Wen Intelligence Group, UCL

Mahdi Dehghan, Shahid Beheshti University

Acknowledgements

TBA

Contact

If you have any questions, do not hesitate to contact us by h.rahmani@ucl.ac.uk or rahmanidashti@gmail.com, we will be happy to assist.