With the growing popularity of "big data", the potential value of
personal data has attracted more and more attention. Applications built
on personal data can create tremendous social and economic benefits.
Meanwhile, they bring serious threats to individual privacy. The
extensive collection, analysis and transaction of personal data make it
difficult for an individual to keep the privacy safe. People now show
more concerns about privacy than ever before. How to make a balance
between the exploitation of personal information and the protection of
individual privacy has become an urgent issue.
In this book, the authors use methodologies from economics, especially
game theory, to investigate solutions to the balance issue. They
investigate the strategies of stakeholders involved in the use of
personal data, and try to find the equilibrium.
The book proposes a user-role based methodology to investigate the
privacy issues in data mining, identifying four different types of
users, i.e. four user roles, involved in data mining applications. For
each user role, the authors discuss its privacy concerns and the
strategies that it can adopt to solve the privacy problems.
The book also proposes a simple game model to analyze the interactions
among data provider, data collector and data miner. By solving the
equilibria of the proposed game, readers can get useful guidance on how
to deal with the trade-off between privacy and data utility. Moreover,
to elaborate the analysis on data collector's strategies, the authors
propose a contract model and a multi-armed bandit model respectively.
The authors discuss how the owners of data (e.g. an individual or a data
miner) deal with the trade-off between privacy and utility in data
mining. Specifically, they study users' strategies in collaborative
filtering based recommendation system and distributed classification
system. They built game models to formulate the interactions among data
owners, and propose learning algorithms to find the equilibria.