Grundbegriffe der Anonymisierung personenbezogener Daten

Many organisations collect data about their processes, customers, the use of their products, and many other topics in order to analyse these data in the context of data mining, big data, machine learning, and similar approaches. However, in most cases such data refer to individual people, and the persons concerned rightly expect that their data are protected adequately and kept private. The resulting limitations on the use of the data often lead to conflicts with and limitations of the data analysis. A technique that helps to overcome these conflicts and limitations is the anonymisation of the data, modifying the data in such a way that they no longer refer to individuals but still allow certain forms of analysis. However, anonymising data turns out to be far more complex than just removing names and other identifiers, and there are many examples where apparently anonymised data were de-anonymised and could be assigned to the individuals concerned after all. Therefore, a number of systematic techniques for evaluating and achieving anonymity, such as k-anonymity and differential privacy, have been developed for this purpose. The current report therefore gives a first overview of the concept of anonymisation, the remaining threats to anonymity, and the main approaches used for anonymising data. The paper concludes with a summary of open research questions for further work.