Dada: Debiased Android Malware DAtasets

The Dada dataset is associated with the paper “Debiasing Android Malware Datasets: How can I trust your results if your dataset is biased?”. The goal of this dataset is to provide a new updated dataset of goodware/malware applications that can be used by other researchers for performing experiments, for example, detection or classification algorithms. The dataset contains the applications hashes and some characteristics. The dataset DOES NOT contain the malware themselves but one can download them with their hash from well-known repositories such as AndroZoo and VirusShare.We provide well-known old datasets (Drebin, AMD) and several extracts of the AndroZoo and VirusShare repository of applications. We also provide the pre-computed output of the debiasing process of labeled biased datasets modified to resemble an extract of AndroZoo. Researchers can directly use these outputs to perform their own experiments. We also provide the scripts that implement the proposed debiasing algorithm to make our experiments fully reproducible.