EEXCESS or the challenge of privacy-preserving quality recommendations (Benjamin Habegger, Nadia Bennani, Elöd Egyed szigmond, Omar Hasan)
Abstract
EEXCESS (eexcess.eu) is a European project willing to improve the visibility of quality contents found in the long tail of the web and hidden by massive content providers such as Google, Wikipedia, etc. Within EEXCESS the envisioned approach consists in identifying the suitable content for a given user by building a specific and precise user profile. While, the main challenge is then to match the specific content to the appropriate users, a crucial induced challenge is that of privacy. Indeed, the collection and enrichment of detailed user profiles requires finding the right compromise between quality recommendations and user privacy. In this talk, we focus on the privacy challenges raised by the EEXCESS project and the directions we are leading privacy related research. In particular, through a tour of an initial prototype, we discuss how we envisage to provide user with a system which, at the same time, (1) provides transparency on the collected private data, (2) empowers users with control on private data disclosure, and (3) provides feedback on the utility of such disclosure. Transparency is provided through user interfaces showing the exactly what private data is being collected by the system. Control is provided by empowering users with a privacy configuration tool alloowing them to control on the levels of disclosure of different privacy-sensitive attributes. Feedback is provided by immediately showing the user the impacts of the privacy settings on the results given by the system. In the EEXCESSS case, results are recommendations for which the user can himself judge of the increase or reduction of quality.
A short bio
Benjamin Habegger has a dual experience both as a researcher and in the software industry, in particular as an entrepreneur. He currently is mainly working with the LIRIS laboratory as a contractual research on the EEXCESS European project. Previously, he has worked in multiple functions of the software industry among which co-founder and chief technical officer of a startup, software architecture, freelance, pre-sales and support. Benjamin Habegger obtained a Phd from the University of Nantes in December 2004. His scientific and technical domains of expertise are information retrieval, data integration, data analytics, knowledge management and machine learning.
