Abstract | ||
---|---|---|
Social media represent one of the most popular online tools to spread and exchange formal and informal information based on specific human interests, habits and purposes. As they are free, easy to use and widely adopted, criminals commonly exploit them to quickly disseminate information and propaganda, to recruit people and so on. Due to the vast usage and breadth of topics discussed on them, it is not trivial to identify criminals abusing of social media for their purposes. Machine learning techniques have already shown benefits in classification problems in different application domains. As a consequence, a trained classifier for the identification of potential malicious users on these platforms would represent a desirable solution. In this perspective, this work explores the possibility of using data which are extracted from public personal features in order to identify potential terrorists on social media, by showing current limits. To this aim, a public dataset on known terrorist's details is combined with a manually collected dataset of public Facebook profiles. The adopted approach is presented and then the data-related issues, which emerged from this experience, are discussed1.
|
Year | DOI | Venue |
---|---|---|
2019 | 10.1145/3339252.3341483 | Proceedings of the 14th International Conference on Availability, Reliability and Security |
Keywords | DocType | ISBN |
Cyber-criminality, Interned-based Crimes, Machine Learning, Social Media Analysis, Terrorist Networks | Conference | 978-1-4503-7164-3 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Andrea Tundis | 1 | 47 | 14.38 |
Leon Böck | 2 | 0 | 0.34 |
Victoria Stanilescu | 3 | 0 | 0.34 |
Max Mühlhäuser | 4 | 1652 | 252.87 |