Logo
latest

Adversaries

  • Advertising eco-systems
  • Black markets
  • Data brokers
  • Data scientists
  • EU Regulators
  • Employers
  • Gray hats
  • Insurance companies
  • Law enforcement
  • Marketers and advertisers
  • Politicians

Assets

  • Auxiliary information
  • Data releases
  • Target dataset

Attack vectors

  • Classification analysis
  • Graph matching
  • Matching features
  • Sparsity-based
  • Trail re-identification

Attacks

  • Inference attacks
  • Linkage attacks
  • Structural attacks

Threats

  • Content disclosure
  • Identity disclosure
  • Link disclosure
  • New form of consent

Assistive technologies

  • Adversary skills
  • Behavioural analysis
  • Content analysis
  • Data storage
  • Link prediction
  • Predictive analysis
  • Spammer elimination
  • Timeliness

Uses

  • Advertising ecosystems
  • Click fraud (wars)
  • Customer churn analysis
  • Customer experience analysis
  • Fraud detection
  • Market analysis
  • Network monitoring
  • Recommendation engines
  • Research & development
  • Risk modelling
  • Sentiment analysis
  • Social network analysis

Impacts

  • ↑ Bias and discrimination
  • ↓ Competition
  • ↑ Data
  • ↑ Illegal and unethical mining
  • ↑ Regulation
  • ↑ Surveillance and tracking
De-anonymisation threat model
  • De-anonymisation threat model
  • Green Team
  • Improbability Blog
  • About
  • Register

Link prediction

Link prediction is a method that an adversary can use to bridge between auxiliary information and a target dataset if they are from different sources and have little in common for matching.

Resources

  • Link Prediction by De-anonymization: How We Won the Kaggle Social Network Challenge, Arvind Narayanan, Elaine Shi, Benjamin Rubinstein, 2011

Previous Next

Unseen University, 2023, with a forest garden fostered by /ut7.
Read the Docs v: latest
Versions
latest
Downloads
On Read the Docs
Project Home
Builds