• ASE Project Proposals
• Abraham Campbell
• Alexey Lastovetsky
• Aonghus Lawlor
• Brett Becker
• Chris Bleakley
• Eleni Mangina
• Fatemeh Golpayegani
• Félix Balado
• Gavin McArdle
• Guenole Silvestre
• Liliana Pasquale
  • Forensic-Ready Systems
  • Adaptive Privacy Management
• Mel Ó Cinnéide
• Neil Hurley
• Performance Engineering
• Rem Collier
• Tony Veale
• Vivek Nallur

Adaptive Privacy Management in Online Social Networks

Despite their widespread adoption, online social networks (OSNs) still provide primitive mechanisms for privacy management: users can only select contacts with which specific contents should be shared.
Shared information can assume varying semantics depending on its content, which, for example, can refer to activities, attended events, physical locations, and people belonging to a user’s social circles. Depending on its semantics, shared information can cause privacy breaches when shared with a specific group of recipients. For example, if shared information refers to a job post, it might be preferred only to include people interested in similar topics or having a professional relationship with the user. In the case of an information related to a party attendance, relevant groups might include people who are physically close, among which a user can remove those who have not been invited.

This project addresses the problem of helping OSN users manage their privacy by recommending an adequate audience “on-the-fly”, as soon as a new content is going to be published. The project will build on the intuition that similarity measures between shared information and groups of contacts can provide useful insights on the selection of appropriate/inappropriate information audiences. Information can grouped depending on type of activity or subject, physical locations, and people involved. Contacts can be grouped depending on thematic subjects of interest, physical locations and social relationships.

The student is expected to implement 3 algorithms to a) compute subject or activity of shared information, b) group contacts and c) compute similarity between shared information and potential recipients. To address this aspect the student is expected to use existing OSN data-sets available publicly.
The student is also expected to create a plugin for an online social network of his/her choice, that suggests a potential audience for the information to be shared before the content is shared. This tool will support management adaptively since groups of contacts can evolve dynamically. In particular, contacts can be added or removed, physical location of users can vary, and interaction activities (e.g., liking, commenting) and social relationships between a user and her contacts can evolve.