Clemens Heidinger, Klemens Böhm, Erik Buchmann, Kai Richter

Karlsruhe Reports in Informatics

In past years, many anonymization schemes, anonymity notions, and anonymity measures have been proposed.
When designing information systems that feature anonymity, choosing a good approach is a very important design choice. While experiments comparing such approaches are enlightening, carrying out such experiments is a complex task and is labor-intensive. To address this issue, we propose the framework FACTS for the experimental evaluation of anonymization schemes.
It lets researchers implement their approaches against interfaces and other standardizations that we have devised. Users can then define benchmark suites that refer to those implementations. FACTS gives way to comparability, and it includes many useful features, e.g., easy sharing and reproduction of experiments. We evaluate FACTS (a) by specifyingand executing a comprehensive benchmark suite for data publishing and (b) by means of a user study. Core results are that FACTS is useful for a broad range of scenarios, that it allows to compare approaches with ease, and that it lets users share and reproduce experiments.

Download Pdf