Masterarbeit
Supporting the Modeling and Visualization of AI Components in Web-BasedSystem Architectures using WAM
Completion
2025/12
Research Area
Students
Ahmad Hijazi
Advisers
Abubaker Gaber
Prof. Dr.-Ing. Martin Gaedke
Description
Artificial Intelligence (AI) is becoming an important part of many modern web-based systems. However, current architecture modeling tools often do not support the specific needs of AI components. These tools lack the ability to clearly represent elements like training processes, inference workflows, datasets, or trust-related information. As a result, it is difficult for developers, designers, and compliance officers to show how AI is integrated into a system in a way that is transparent, understandable, and aligned with legal or ethical standards.
The goal of this thesis is to extend the existing modeling tool Wamflow, which is based on the Web Architecture Model (WAM). While Wamflow supports basic system elements such as services, applications, and process, it currently does not support AIspecific elements or semantic metadata. To address this, the tool will be extended by adding four new visual elements: AI Service, AI Process, AI Dataset, and AI Application. These elements will have built-in fields for describing key AI properties such as data source, accuracy, model type, and compliance tags. In addition, the tool will support export in structured formats like RDF and JSON, allowing models to be reused for documentation, validation, or auditing.To evaluate the solution, the project will follow a mixed-methods approach. A user study will be conducted where participants use the extended Wamflow to complete a modeling task. Their performance, accuracy, and feedback will be collected through standard usability questionnaires and short interviews. At the same time, the final version of the tool will be compared to a list of defined functional and non-functional requirements. These include ease of use, support for metadata, extensibility, and integration capabilities. The results of this evaluation will show how well the extended Wamflow supports the modeling of AI-based systems in practice.
