LinkMirage: Enabling Privacy-preserving Analytics on Social - PowerPoint PPT Presentation

Outline Introduction LinkMirage Conclusion LinkMirage: Enabling Privacy-preserving Analytics on Social Relationships Changchang Liu, Prateek Mittal Email: cl12@princeton.edu, pmittal@princeton.edu Princeton University February 23, 2016 1 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion Social relationships (a) (b) Third party applications rely on users’ social relationships: • E-commerce • Spam detection • Anonymous communication • Sybil defenses 2 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion Social relationships are very sensitive! Social relationships represent • Trusted friendships • Important interactions • Even more, business relations, etc. 3 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion How to balance utility and privacy? Privacy Utility Protect privacy of sensitive social relationships Preserve utility of obfuscated social relationships for real-world applications 4 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion Previous work of link privacy mechanisms To protect link privacy, previous work • obfuscate social relationships through link additions/deletions G G � p p add del However, previous work • only consider graph data where the links are redstatic 5 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion Limitations of previous link privacy mechanisms To protect link privacy, previous work • obfuscate social relationships through link additions/deletions G G � p p add del However, previous work • only consider graph data where the links are static 5 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion However, social networks are dynamic Temporal Facebook dataset (every three months) with 46,952 users and 876,993 edges 6 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion However, social networks are dynamic An adversary can combine the previously perturbed graphs together obfuscation G ¢ G 0 0 ... ... obfuscation G - ¢ G - t 1 t 1 obfuscation G G ¢ t t Adversary 7 / 18

Outline Introduction Social Relationships LinkMirage Privacy-utility tradeoff Conclusion Our Objective • Balance privacy and utility • Handle both the static and dynamic social network topologies • Provide rigorous privacy guarantees • Useful in real-world applications 8 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis LinkMirage LinkMirage Overview Algorithm Description Privacy Analysis Utility Analysis 8 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Social Relationship based Applications Q Original Graph Untrusted ( ) G Q G ¢ Applications Privacy-preserving graph analysis Sybil defenses OSN Anonymous communication providers ... 9 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Privacy-preserving Social Relationship based Applications Q Original Graph Untrusted ( ) G Q G ¢ Applications Privacy-preserving graph analysis Sybil defenses OSN Anonymous communication providers ... 9 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis LinkMirage Architecture LinkMirage System (Trusted) Obfuscation Algorithm Q Original Graph Obfuscated Graph Untrusted G ¢ ( ) G Q G ¢ Applications Privacy-preserving graph analysis Sybil defenses OSN LinkMirage Anonymous communication social link app providers User1 User2 User3 � ... 9 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis LinkMirage LinkMirage Overview Algorithm Description Privacy Analysis Utility Analysis 9 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Key intuitions • Naive method: independent perturbation − more information is leaked to others • We need to − incorporate graph evolution − leverage the information already released in previous graphs 10 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description G - t 1 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description G - t 1 1. clustering C 1 C 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description G - t 1 1. clustering C 1 C 2 2. perturbation ¢ G - C ¢ 1 t 1 C ¢ 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description 3. evolution G - G t 1 t 1. clustering C 1 C 2 2. perturbation ¢ G - C ¢ 1 t 1 C ¢ 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description 3. evolution G - G t 1 t 1. clustering 4. dynamic clustering C C 1 1 C 3 C C 2 2 2. perturbation ¢ G - C ¢ 1 t 1 C ¢ 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description 3. evolution G - G t 1 t 1. clustering 4. dynamic clustering C C 1 1 C 3 C C 2 2 2. perturbation 5. selective perturbation ¢ C ¢ C ¢ G ¢ G - C ¢ 3 1 1 t 1 t C ¢ C ¢ 2 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Algorithm Description 3. evolution G - G t 1 t 1. clustering Key step 1: dynamic clustering C C 1 1 C 3 C C 2 2 2. perturbation Key step 2. selective perturbation ¢ C ¢ C ¢ G ¢ G - C ¢ 3 1 1 t 1 t C ¢ C ¢ 2 2 11 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Two Key Steps in Our Algorithm Two key steps • Dynamic Clustering – find communities by simultaneously considering consecutive graphs – backtrack based on clustering result of the previous graph • Selective Perturbation – perturb the minimal amount of edges – use a very high privacy parameter while preserving structural properties (utility) 12 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Facebook Temporal Dataset (46,952 users and 876,993 edges) Original graphs t=3 t=4 t=5 Superior utility, due to dynamic clustering Utility advantage even exists in static scenario 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Utility Advantage Mittal et al. LinkMirage Original graphs t=3 t=4 t=5 Superior utility, due to dynamic clustering Utility advantage even exists in static scenario 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Utility Advantage Mittal et al. LinkMirage Original graphs t=3 t=4 t=5 Superior utility, due to dynamic clustering Utility advantage even exists in static scenario 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Privacy Advantage Original graphs Overlapped edges (black) and Changed edges (yellow) between consecutive graphs Superior privacy, due to selective perturbation 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Privacy Advantage Original graphs Mittal et al. LinkMirage Overlapped edges (black) and Changed edges (yellow) between consecutive graphs Superior privacy, due to selective perturbation 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis Privacy Advantage Original graphs Mittal et al. LinkMirage Overlapped edges (black) and Changed edges (yellow) between consecutive graphs Superior privacy, due to selective perturbation 13 / 18

Outline LinkMirage Overview Introduction Algorithm Description LinkMirage Privacy Analysis Conclusion Utility Analysis LinkMirage LinkMirage Overview Algorithm Description Privacy Analysis Utility Analysis 13 / 18