.svg.png)
Geo-Semantic User Profiling & Privacy
This research programme investigates how location data from mobile devices and social networks can be used to build rich semantic profiles of users — and critically, how to protect user privacy when such profiles can reveal sensitive personal information.
Spanning from 2011 to 2017, four PhD projects progressively developed methods to extract meaningful profiles from location traces, enrich them with place semantics, model their temporal dynamics, and ultimately address the privacy implications of location disclosure on geo-social networks.
The research follows a natural progression:
Geographical Places as a Personalisation Element: Pioneered the extraction of semantically rich profiles from GPS mobility logs using place affordance — the typical human activities and services provided at certain place types. Introduced Personal Gazetteers and the PGML markup language for encoding spatial, temporal, and thematic user information.
Extracting Place Semantics from Geo-Folksonomies: Developed a framework for analysing geo-folksonomies (collaborative place tagging) to discover place-related semantics. Built place type and activity ontologies from social tagging data, and proposed user similarity and place similarity measures based on the extracted semantics.
Spatiotemporal User and Place Modelling on the Geo-Social Web: Proposed static and dynamic user models capturing how user preferences change over time. Developed semantic place models identifying user activities and relationships between places, users, and activities. Created novel location and content recommendation methods using Foursquare data.
Location Privacy Awareness on Geo-Social Networks: Addressed the critical privacy implications of location sharing. Developed a privacy threat model considering shared data types, visibility by adversaries, and user awareness. Proposed privacy feedback solutions that reveal risk levels and enable users to make informed decisions about location disclosure.
This research programme established foundational methods for understanding how location data reveals information about users, their interests, and their activities. The privacy-focused culmination ensures that as location-based services proliferate, users can be empowered to understand and control their location disclosure.