Supporting the Identification and the Assessment of Suspicious Users on Twitter Social Media

Abstract

The exploitation of Internet technology represents for terrorists and criminals a convenient means for establishing and advertising illegal activities. Especially, social networks facilitate new collaborations as well as the spreading of information with a lower risk of being exposed and fetched. Indeed, due to the increasing number of social media and the huge amount of data continuously generated from them, the discovering process of cyber-criminals is a hard task to be performed by the Law Enforcement Agencies and Police Forces if only based on traditional approaches. It becomes even harder if the heterogeneous nature of data, due to multi-cultural aspects, such as the variety of languages, is considered during the searching process. As a consequence, the adoption of a computer-based approach represents a viable solution. In particular, this paper aims at supporting the automatic identification process of potential online suspicious users, who act on social media. A methodological process, centered on the combination of well-known text analysis techniques by considering multi-language aspects, is proposed. In addition, an evaluation approach, based on the exploitation of different qualitative evaluation criteria, is employed to assess the level of suspiciousness of the identified users. Finally, a software tool that supports the execution of the proposed process is developed and its experimentation is shown through a case study on Twitter.