Privacy preservation using multi-context systems and default logic

Preserving the privacy of sensitive data is one of the major challenges the information society has to face. Traditional approaches focused on infrastructures for identifying data which is to be kept private and for managing access rights to these data. However, although these efforts are useful, they do not address an important aspect: While the sensitive data itself can be protected nicely using these mechanisms, related data, which is deemed insensitive per se, may be used to infer sensitive data. This inference can be achieved by combining insensitive data or by exploiting specific background knowledge of the domain of discourse. In this paper, we present a general formalization of this problem and two particular instantiations of it. The first supports query answering by means of multi-context systems and hybrid knowledge bases, while the second allows for query answering by using default logic.