Authority-Dependent Risk Identification and Analysis in online Networks

The ADRIAN sub-project aims to automatically monitor selected (running) apps and analyze the collected data, correlate it with social media profiles and form clusters of people in order to identify potential targets and assess their risk potential. If this information is correlated with other classified materials, a risk plausibility can be determined for the corresponding persons (groups) or locations.

Tasks and objectives

Semantic networking on the web, which has been progressing for years, has created a huge, freely accessible source of information for a variety of data-driven applications, which can pose a personal risk under certain circumstances. User-generated data (so-called “user-generated content”) is being linked automatically with existing resources (so-called knowledge sources) more and more effectively. In this way, even seemingly trivial and sometimes unintentionally disclosed individual pieces of information can have harmful consequences for individuals, professional groups or entire institutions. In particular, the linking of social media accounts and posts (e.g. Twitter or Instagram) with movement profiles and location data from popular running apps makes users and their relatives identifiable, traceable and potentially the target of online attacks. The fact that military locations can be localized using the shared geodata from running routes is another security-relevant aspect in this context.

As part of this project, selected running apps will first be monitored and the geodata collected will then be analyzed. The user profiles of running apps and social media platforms are then correlated in order to form a cluster of people and enable the identification of potential targets. As a so-called “digital (running) twin” can be reconstructed in this way as part of the data analysis and knowledge acquisition, extremely sensitive data is generated. If this data can also be correlated with other confidential data (e.g. from security authorities or military agencies), it is possible to assess the plausibility of the threat to the relevant persons (groups) or locations. In order to achieve these goals, the technical implementation of the project must combine methods of information retrieval with approaches from forensic linguistics. Furthermore, methods for network analysis and clustering are used to develop new evaluation functions for the assessment of endangered targets (persons, locations, etc.) on the basis of the information disclosed in Web 2.0. The use of highly secure quantum encryption is also planned for the subsequent transmission of the knowledge gained to other services.