Towards Early Detection of Algorithmic Bias from Dataset's Bias Symptoms: An Empirical Study

This repository contains the appendix and the replication package for the paper Towards Early Detection of Algorithmic Bias from Dataset's Bias Symptoms: An Empirical Study.

The repository is organized as follows:

•  DATASETS.md: contains the list of datasets used to create the bias symptoms dataset, along with additional statistics and references.
•  replication_package: contains the replication package for the analyses reported in the paper. Refer to the [README](replication_package/README.md) file in the replication_package folder for more details.

Data Availability

The bias symptoms datasets used in the experiments are available in the replication_package/data folder. For space constraints the original 24 datasets used to create the bias symptoms dataset are not included in this repository. However, the datasets can be download from the references reported in the `DATASETS.md` file and preprocessed by one-hot encoding categorical values and removing missing values.

Citation

@article{d5143265towards,
  title={Towards Early Detection of Algorithmic Bias from Dataset's Bias Symptoms: An Empirical Study},
  author={d'Aloisio, Giordano and Di Sipio, Claudio and Di Marco, Antinisca and Di Ruscio, Davide},
  journal={Available at SSRN 5143265}
}