Hiding Co-occurring Sensitive Patterns in Progressive Databases

摘要

These days lot of work is been carried out in the field of privacy preserving data mining. Apart from the standard techniques of privacy preservation, methods are being proposed specific to the data mining tasks carried out. However most of these methods work on static databases. One such a method is downgrading application effectiveness. The effectiveness of applications may be downgraded by hiding sensitive association rules, hiding sensitive sets of patterns etc. Sometimes although a particular pattern is not interesting, its co-occurrence with another pattern may reveal certain sensitive information. In this paper we present a novel technique to hide sensitive co-occurring sequential patterns. The proposed method works on progressive databases. Progressive databases are a generalized model of static, dynamic and incremental databases. The applicability of the method is also extended to suit these different types of databases.