• correlation
• risk assessment
• cost-sensitive classifiers
• social network data
• degradation
• analysis
• privacy risk
• social network
• data mining
• classification

We propose a study of the privacy risk on social network data using the state-of-the-art framework. We perform direct simulation of different privacy attacks to compute privacy risk. We also show the results of privacy risk distributions based on the type of privacy attack and realize degradation analysis on the social network graph. To tackle the high computational complexity of such simulation we use an alternative data mining approach to estimate privacy risk using node level metrics. In the end, we compare the computational complexities of two approaches and evaluate results.