Internship subject:
Privacy protection in log analysis

Name of the host laboratory: 
LIX, Ecole Polytechnique

Supervisors:  
Kostas Chatzikokolakis and Catuscia Palamidessi (INRIA and LIX)
Kreshnik Musaraj (Thales)

Internship description:
This stage will be part of a project involving the Laboratory of Computer Science of the Ecole Polytechnique and the industrial group Thales. The project aims at developing a system to provide privacy protection in log analysis. The goal is to allow the analyst to detect from logs suspicious and possibly foul activity, while protecting at the same time the privacy of the employees who have originated those logs.

The idea is to use an aggregation-oriented database technology tike OLAP [1] to represent the data of interest and their correlation, and then to apply differential privacy, a framework recently emerged in the community of statistical databases, to ensure the protection of sensitive and personal information. The typical way this protection is achieved is by adding some controlled noise to the answer, every time the database is queried.

The intern will have the opportunity to work both at the practical level, to help with concrete case studies and the development of the system and its testing, and at the conceptual level, to ensure that the intended degree of privacy is met.

The intern will receive an internship salary financed by Thales.

The internship is supposed to be used by the candidate to produce his/her master thesis. There will be the possibility of continuing working with the team, on the topic of Differential Privacy,  as a PhD student. A preference will be given to the students who are willing to continue as PhDs. 

Bibliography: 
[1] Gray, Jim; Bosworth, Adam; Layman, Andrew; Priahesh, Hamid. "Data Cube: A Relational Aggregation Operator Generalizing Group-By, Cross-Tab, and Sub-Totals". Proc. 12th International Conference on Data Engineering. IEEE. pp. 152–159

[2] Cynthia Dwork. Differential Privacy. International Colloquium on Automata, Languages and Programming (ICALP 2006), p. 1–12.


Contact people:
Kostas Chatzikokolakis and Catuscia Palamidessi

Team:
Comète (http://www.lix.polytechnique.fr/comete/index.php)

Candidate’s required level:
Master’s student (2nd year).  

Proposed Duration:
4-6 months

Prerequisites:
Attitude for formal and mathematical reasoning
Some knowledge of Probability Theory and/or Database Management Systems would be welcome