ABSTRACT
The previous four chapters discussed various aspects of privacy on social networks. Chapter 14 discussed our approach to analyzing social networks. Chapter 15 discussed a new classification method that takes advantage of link types. We showed that by intelligently weighting those link types, it is possible to gain even more benefit from the link types. In Chapter 16, we created artificial links between regional influencers and those who follow in their wake. In Chapter 17, we showed that by intelligently examining historical social media data, we are able to partition the full data into smaller sets, which both increases classification accuracy and reduces the time required for classification. These approaches have given us an understanding of social network analysis and of their privacy implications so that we can develop privacy-enhanced social network analysis techniques. In this chapter, we will add a new dimension to our work in social network analysis. In particular, we show that even if we hide various links and details in social networks, one can still predict various sensitive and private data.
