Why Open Data May Threaten Your Privacy Sebastian Pape, Jetzabel Serna-Olvera, Welderufael B. Tesfay Goethe University Frankfurt Chair of Mobile Business and Multilateral Security Workshop on Privacy and Inference September 21st, 2015 Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 1 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Overview 1 Open Data / De-Anonymization 2 Proposed Approach 3 Conclusion Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 2 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Open Data / De-Anonymization Open Data De-Anonymization Broad range of Applications Often works by linking data sets “unexpectedly” Services helpful to society (e.g. health, educational services) Gets easier with more Open Data Balancing act between usefulness and anonymization Machine Learning allows to work with fuzzy data Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 3 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Tool Support Not the fault of anonymization algorithms Tool support to identify relevant Open Data needed Several capabilities for machine learning approaches Scope limited to Open Data Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 4 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Mirroring Privacy-Related Open Data Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 5 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Referencing Privacy-Related Open Data Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 6 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Comparison Mirroring Referencing + Usability - Usability = Quality of Prediction + Quality of Prediction + Versioning - Versioning = Updates - Updates - Storage + Storage = Bandwidth - Bandwidth Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 7 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Proposed Steps Steps C Collection (O) L Linkage (T) U User Interaction (O/T) A Anonymization as a Service (T) D De-Anonymization-Tests as a Service (T) Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 8 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Challenges Steps Challenges C1 Rate Privacy-Relevance C2 Version Control System L1 Context of Database C Collection (O) L2 Field Names L Linkage (T) DCAT, VoID U User Interaction (O/T) L3 Sparse matches A Anonymization as a Service (T) U1 Structure vs. Full Data D De-Anonymization-Tests as a A1 Server Side vs. Client Side Service (T) Analysis D1 Deterministic vs. Probabilistic Model Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 9 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Conclusion / Discussion Open Data: More attention should be paid to privacy risks Tool support needs to be improved Publication of Open Data should not be prevented Ohm (2009) vs. O’Hara (2011) Useful tool or threat? → Short-term: threat → Long-term: useful tool Should the tool regard leaked/stolen data? Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 10 / 11
Open Data / De-Anonymization Proposed Approach Conclusion Feedback Feedback sebastian.pape jetzabel.serna welderufael.tesfay @m-chair.de Pape, Serna-Olvera, Tesfay Why Open Data May Threaten Your Privacy 11 / 11
Recommend
More recommend