Unternehmensmodellierung zur Vermeidung von Diskriminierung beim Einsatz von Algorithmen

Bachelor Thesis Business Information Systems


    It is now well known that (machine learning) algorithms have discrimination potential. As their influence continues to grow, companies are increasingly responsible for ensuring that their algorithms comply with legal requirements, such as those of the AGG. However, the complexity of algorithmic systems - especially machine learning - makes this difficult. Suitable methods are needed that not only help developers to prevent discrimination in algorithms. Enterprise modeling will be examined in this thesis for its potential to support these problems. To do this, the terms algorithm, machine learning algorithm, and discrimination must be subjected to analysis. The findings from the analysis will be captured in a modeling method, which finally includes a procedure model for the development of „fair“ algorithms.