| Kurzfassung
|
Modern software systems process large amounts of personal data, making it essential to document data flows in diagrams to comply with regulations such as the GDPR. The tool xDECAF allows users to graphically create and analyze such diagrams and identify constraint violations, but large technical diagrams remain difficult for experts from non-technical disciplines to understand. The approach presented in this thesis translates constraint violations into natural language by first extracting relevant data in tabular form and then passing it to a large language model (LLM). The LLM then generates a natural-language description of the violated constraint along with an explanation of how the violation occurred. A two-stage online survey provides initial evidence that natural-language summaries can positively influence the comprehension of violations and that the implemented prototype offers good usability.
|
