Personalized anonymity in social networks data publication

Abstract

Social networks consist of entities connected by links representing relations. Social networks applications have become popular for sharing information. Many social networks contain highly sensitive data. So some privacy preservation technologies are already proposed in social networks data publication. However, the existing technologies focus on a universal approach that exerts the same level of preservation for all entities, without catering for their concrete needs. Motivated by this, we present a k-neighborhood anonymous method based on the concept of personalized anonymity. We divide entities into sensitive and non-sensitive. The entities declare their publication requests when submitting data. Our technique performs the minimum modification on origin social networks for satisfying every entity's neighborhood privacy requirement and retains the largest amount of information from the published networks. We develop an algorithm against 1-neighborhood attack and execute experiments on the synthetic dataset to study the utility and publication quality.