A K-anonymity Based Semantic Model For Protecting Personal Information and Privacy

The proper protection of personal information is increasingly becoming an important issue in an age where misuse of personal information and identity theft are widespread. At times there is a need however for management or statistical purposes based on personal information in aggregated form. The k-anonymization technique has been developed to de-associate sensitive attributes and anonymise the information needed to a point where the identity and associated details cannot be reconstructed. The protection of personal information has manifested itself in various forms, ranging from legislation, to policies such as P3P and also information systems such as Hippocratic database. Unfortunately, none of these provide support for statistical data research and analysis. The traditional k-anonymity technique proposes a solution to this problem, but determining which information can be generalized and which information needs to be suppressed is potentially difficult to determine. In this paper we propose a new idea that integrates personal information ontology with the concept of k-anonymity, in order to overcome these problems. We demonstrate the idea with a prototype in the context of healthcare data management, a sector in which maintaining the privacy of individual information is essential.