Database anonymization

摘要

At least one quasi-identifier attribute of a plurality of ranked attributes is selected for use in anonymizing a database. Each of the ranked attributes is ranked according to that attribute's effect on a database-centric application (DCA) being tested. In an embodiment, the selected quasi-identifier attribute(s) has the least effect on the DCA. The database is anonymized based on the selected quasi-identifier attribute(s) to provide a partially-anonymized database, which may then be provided to a testing entity for use in testing the DCA. In an embodiment, during execution of the DCA, instances of database queries are captured and analyzed to identify a plurality of attributes from the database and, for each such attribute identified, the effect of the attribute on the DCA is quantified. In this manner, databases can be selectively anonymized in order to balance the requirements of data privacy against the utility of the data for testing purposes.

主权项

1. A method for database anonymization, the method comprising:
selecting a quasi-identifier attribute from ranked attributes to provide a selected quasi-identifier attribute, wherein each of the ranked attributes is ranked according to a number of statements in code implementing a database-centric application affected by the ranked attribute; and anonymizing, by a processor, a database based on the selected quasi-identifier attribute to provide a partially anonymized database.