Description
In our BMBF-funded project Privacy E2E we are working on privacy preserving AI-systems. In the context of this project, we explore domain-specific languages (DSL) to express privacy requirements and assess the privacy implications of various components within AI-enabled systems.
This thesis offers an opportunity to design a DSL aimed at annotating structural and behavioral models of AI systems. The goal is to enable precise identification and management of sensitive data across diverse use cases and within specific AI pipeline modules. The work will focus on creating a language that helps embed the principle of data minimization both in system design and during runtime, supporting static and dynamic privacy verification for AI systems.
Previous work has identified different data categories and developed an ontology, which will form the basis of this DSL. Ultimately, it will provide a framework for ensuring that privacy requirements are considered at every stage of AI system development.
Prerequisites: Knowledge on domain-specific languages (DSL), e.g. from our course Software Languages
Contact: Yorick Sens (yorick.sens@rub.de)
Extent: M.Sc.