Topic area: Modeling
The ENCASE project aims to leverage the latest advances in web security and privacy to design and implement a browser-based architecture for the protection of minors from malicious actors in online social networks, by exploiting sentiment and affective analysis along with graph mining.
The ENCASE user-centric architecture consists of three distinct services, which can be combined to customize protection of vulnerable groups (i.e. youngsters) from cyberbullying, sexual abuse, identity theft and phishing attacks. The three basic components with their respective back-end software stack are:
The component that collects the users’ online actions to unveil incidents of aggressive or distressed behavior
The analytic component that mines social web data to detect fraudulent and fake activity and raises alerts to the user
The detection component that alerts the user when she/he is about to share sensitive content with inappropriate audience online and raises privacy threat
The main goal is to utilize openly available information from the web through browser sessions of the user and profile associated users in her/his network to determine whether they could pose a threat. This is performed through user experience assessment, large scale data processing and complex network analysis to create features that characterize user behavior online. As a result, new potential threats are identified in the web such as users that help link criminals to the rest of the network (social bridges) or a detailed distinction between bullies, aggressors and harassers. In the next step, machine learning and predictive modeling are employed to classify users based on the aforementioned features along with content confidentiality techniques that help prevent underage users from sharing sensitive content with malicious users.
So far, extensive analyses and modeling of online user behavior has taken place along with pilot testing and extensive experimentation with use cases to create the system architecture with all the software and data requirements. Promising results after prototype testing indicate that with a combination of complex network and affective analyses threats that exceed individual intuition and understanding can be identified and real time detection and protection of underage users can occur.