Privacy in data mining

Authors views on combining computer science and statistics to foster the development of privacy-preserving data mining (PPDM) are described. In the first paper authors determined which PPDM techniques are best to protect sensitive information, and how the, quality and privacy measures must be defined. The second paper analyzes the problem of confidentiality in categorical statistical databases when association rules are to be preserved. The third paper proposes to use probabilities to define bounded information loss measures for any statistic of interest. The fourth paper deals with k-anonymity, which is a useful concept to manage the conflict between data quality and individual privacy.