Abstract
Nowadays, social networks have gained popularity among the world. Social networks have a lot of data about interaction among entities and these data may contain the individual privacy content. Accordingly, many studies have been proposed to protect the privacy in social networks. The previous works only focus on developing the privacy preserving methods for releasing a single anonymized graph used to represent a social network. However, the single anonymized graph may not be enough for analyzing the evolution of the whole network. Therefore, we address a novel problem of preserving the privacy of interaction among entities for multiple releases on time-series social network data in this thesis, which means we will release multiple anonymized graphs to represent a social network with time-series data. We point out that the privacy may be revealed across the multiple releases, if we apply the existing methods to generate the individual anonymized graph for the network in the different timestamps. We provide an anonymization method for releasing multiple anonymized graphs at one time on time-series social network data. We detail our experiment steps and evaluate the utility of the anonymized graphs by answering a series of aggregate queries. The results show that the multiple releases generated by our method answer the queries accurately.