Computer Security
Technical Committee

[switch to italian]

A widely recognized problem in computer security is the lack of public datasets for a sound evaluation of security solutions. Public datasets can allow for reproducibility of scientific evaluations, but data privacy often represents an obstacle. In this section we link you to the most important (few) datasets of public domain, with the hope of facilitating the scientific evaluation process.

DARPA'99 DatasetThe historical dataset created by DARPA for the evaluation of Intrusion Detection Systems. Since this datasets has been created through artificial procedures, nowadays it is not representative of real traffic. Nevertheless, it has been the first, and is still one of few public datasets that allow the evaluation of Intrusion Detection Systems.
Web DatasetAn anonymized dataset of real web traffic/attacks, that can be downloaded from the PRALab website (University di Cagliari).
DB ForgeryDataset for the experimental evaluation image forgery detection techniques, built by Image Processing Lab (IPLab), Department of Matematics and Computer Science, University of Catania.
Digital is a website of digital corpora for use in computer forensics education research.
Android PRAGUARDThis Android Dataset is composed by roughly 10500 samples that have been obtained by obfuscating existing malware with seven different obfuscation techniques.
UNSW-NB15Dataset composed by more than two million of samples whose information have been extracted by more than 100 GBs of traffic.