Sensitive attribute prediction for social networks users

Younes Abid 1 Abdessamad Imine 1 Michael Rusinowitch 1
1 PESTO - Proof techniques for security protocols
Inria Nancy - Grand Est, LORIA - FM - Department of Formal Methods
Abstract : Social networks are popular means of data sharing but they are vulnerable to privacy breaches. For instance, relating users with similar profiles an entity can predict personal data with high probability. We present SONSAI a tool to help Facebook users to protect their private information from these inferences. The system samples a subnetwork centered on the user, cleanses the collected public data and predicts user sensitive attribute values by leveraging machine learning techniques. Since SONSAI displays the most relevant attributes exploited by each inference, the user can modify them to prevent undesirable inferences. The tool is designed to perform reasonably with the limited resources of a personal computer, by collecting and processing a relatively small relevant part of network data.
Document type :
Conference papers
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01939283
Contributor : Abdessamad Imine <>
Submitted on : Monday, December 31, 2018 - 7:12:10 PM
Last modification on : Thursday, January 3, 2019 - 11:56:15 AM
Long-term archiving on : Monday, April 1, 2019 - 12:57:10 PM

File

WORKSHOP_EDBT.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01939283, version 1

Collections

Citation

Younes Abid, Abdessamad Imine, Michael Rusinowitch. Sensitive attribute prediction for social networks users. DARLI-AP 2018 - 2nd International workshop on Data Analytics solutions for Real-LIfe APplications, Mar 2018, Vienne, Austria. ⟨hal-01939283⟩

Share

Metrics

Record views

60

Files downloads

52