Towards Privacy-Preserving Data Mining in Law Enforcement

For law enforcement to be effective, it needs to extract previously unknown knowledge from large amounts of different types of data. Data mining is the most compelling tool for this task as it is motivated by successful applications in numerous domains. Therefore, many believe that data mining can significantly improve the execution of law enforcement. However, a severe problem occurs when data mining is applied: many inevitable mistakes result in privacy violations. Recently, we developed a new approach to data mining, called the ROC isometrics approach, which is proven to produce reliable outputs in the sense that we can set the number of mistakes before the data mining is actually applied. In the paper, we determine the implications of the approach to law enforcement and we propose several recommendations for legislations that try to deal with data mining. As a result, we may conclude that the ROC isometrics approach allows for privacy-preserving data mining so that law enforcement becomes more effectively and efficiently than so far.