Jump to content
Image picture of a abstract network

Data Science for Public Security

DSAI makes the internet safer for European citizens

The Internet is an unprecedented platform for communication, information and commerce. At the same time, it is an unprecedented platform for disinformation, fraud and criminal activity. The rapid pace of technological change makes regulation and enforcement in this digital realm difficult. DSAI cooperates with public authorities and law enforcement to make the internet safer for citizens and commerce, in three crucial areas: Digital Forensics, Virtual Asset analysis, and Consumer Protection.

Digital Forensics has two aspects: first, the validation and verification of the provenance and authenticity of digital media, and second, the automated extraction of features and events from said media. In the first case, DSAI applies machine learning and natural language processing to the challenge of disinformation detection. In the second case, DSAI applies its unique expertise in audio analysis to extract security-relevant audio events using neural network models.

Virtual Assets - developments in cryptocurrencies beyond Bitcoin - are digital representations of value that can be traded or transferred digitally and used for payment or investment purposes. Virtual assets are facilitating new forms of crime and impose novel forensic challenges for law enforcement and prosecutors of financial crime. DSAI’s GraphSense is a virtual asset analytics platform with an emphasis on full data sovereignty, algorithmic transparency, and scalability. GraphSense is open source and free and provides a dashboard for interactive investigations and, more importantly, full data control for executing advanced analytics tasks.

Consumer Protection: Time of discovery is the most important and critical factor to successfully prevent citizens from being exploited by fraudulent e-commerce offerings. DSAI has collected and published an annotated ground-truth collection of archived Fake-Shops – the first of its kind. In the FFG MAL2 project, DSAI developed machine-learning models that achieve a 97% accuracy in classifying Fake-Shops correctly, solely based on intrinsic features of the site. Through a browser-plugin, customers can receive model-based warnings of unknown threats in real-time while shopping online. Open Source tools such as an Expert Analysis Dashboard or the Fake-Shop Database are used daily at customer protection agencies such as Watchlist Internet and allow for a higher throughput, provide means for semi-automation and interfaces for supporting an integrated Fake-Shop detection lifecycle.



  • GraphSense
  • MAL2
  • preis.wert