A graph based approach for privacy preservation of citizen data in e-governance applications

Online social media provides a platform for the citizens to create profiles and share information. This has also enabled the governments to create their own profiles and promote different e-governance services and policies. It is possible to analyze the responses of citizens available in text form through text mining and infer interesting patterns. Also, governments may ask for citizen profiles and connection structures to strategize decision making. Such analysis on structural information can help governments to identify communities that get benefit from certain government services. Graph is a common way to understand such connections. However, the simple graph model is unable to represent super-dyadic relationship. Hypergraph model is a generalized graph model which can be used in such scenarios to represent complex connections in groups. But, analysis of data irrespective of its representation can raise privacy threats. This paper considers the identity disclosure attack on hypergraph structures. Identity disclosure attacks can disclose the identity of the users leading to disclosure of sensitive information associated with the users. Hence, government must anonymize the data before subjecting it to analysis by any third party. The focus of this paper is hypergraph anonymization where we study a rank-label attack model which is a stronger attack than the existing rank attack model. The results show the vulnerability of rank anonymization against rank-label attack. In the sequel, a Sequential Rank-Label Anonymization (SRLA) approach has been proposed which outperforms the Greedy Rank-Label Anonymization (GRLA).