Privacy protection in personalized web search: A peer group-based approach

Abstract

Privacy protection in web search engines is becoming more and more serious in recent days. In this paper, we study the problem of privacy protection in web search, with a special focus on IP-address based personalized web search. Our goal is to break the linkage between users' identities (e.g., IP address) and their issued queries so as to prevent privacy breaches. Our privacy model, which shares similar characteristics of l-diversity in privacy preserving data publishing of relational data, provides a strong privacy guarantee in web search. The central idea of our privacy model is to protect user's search activities within a social peer group. A social peer group contains a set of individual users. From search engines's perspective, search queries issued by users from the same peer group cannot be uniquely linked to individuals within the group. A framework based on grouping social peer users is proposed to achieve the privacy requirement. We also provide some experimental results to show that our methods achieve high efficiency in practice. © 2013 Springer-Verlag.