Data utility and privacy protection in data publishing

摘要: Data about individuals is being increasingly collected and disseminated for purposes such as business analysis medical research. This has raised some privacy concerns. In response, a number of techniques have been proposed which attempt to transform data prior its release so that sensitive information the contained within it protected. A:-Anonymisation one technique attracted much recent attention from database research community. works by transforming in way each record made identical at least A: 1 other records with respect those attributes are likely be used identify individuals. helps prevent associated disclosed, individual represented dataset. Ideally, /c-anonymised dataset should maximise both utility protection, i.e. allow intended analytic tasks carried out without loss accuracy while preventing disclosure, but these two notions conflicting only trade-off between them can achieved practice. The existing works, however, focus on how either or protection requirement may satisfied, often result anonymised an unnecessarily and/or unacceptably low level protection. this thesis, we study construct /-anonymous satisfies requirements. We propose new criteria capture requirements, algorithms A:-anonymisations required utility/protection guarantees generated. Our extensive experiments using benchmarking synthetic datasets show our methods efficient, produce A:-anonymised desired properties, outperform state art retaining providing