Preserving the Support of Sensitive Item(s) while Hiding Sensitive Association Rules

Abstract

An essential data-mining method for identifying intriguing relationships among a sizable collection of data objects is association rule mining. It may be a threat to the privacy of uncovered confidential information since it may reveal patterns and other types of sensitive knowledge that are hard to obtain in other ways. Such data needs to be shielded against unwanted access. Numerous tactics had been put forth to conceal the knowledge. Some employ data disruption, clustering, data distortion, and distributed databases across multiple sites. The need to strike a balance between the user's legitimate needs and the secrecy of exposed data is a challenge with hiding sensitive rules that has not yet received enough attention. The suggested method makes advantage of the data distortion methodology, which modifies the sensitive elements' position without changing their support. The database is still the same size. It first prunes the rules using the concept of representative rules, and then it conceals the rules that are sensitive. This strategy has the advantage of hiding the maximum number of rules; in contrast, the existing ways are unable to conceal all the needed rules, which should be concealed in the fewest number of passes. The suggested method is also contrasted with current approaches in the paper.