Individual learning paths and adaptive support represent central challenges in modern higher education. Precise assessment of learning status and data-driven estimation of the Zone of Proximal Development (ZPD) form the foundation for effective AI-based scaffolding. Within the scope of the Multi-Agent Scaffolding project at the University of St. Gallen, a profiling agent is to be developed and evaluated both conceptually and practically.

You can flexibly select the focus of your work: (a) empirical requirement analysis involving students, (b) systematic literature review of state-of-the-art AI-based learning diagnostics, or (c) conceptual taxonomy work for classification and comparability of ZPD estimation methods.

• Collecting and analyzing requirements for profiling and diagnostic agents via qualitative or quantitative user surveys.

• Systematic literature review on existing methods for AI-supported learning diagnostics and ZPD estimation.

• Developing a robust taxonomy of ZPD diagnostics and scaffolding approaches, potentially including mapping to international best practices.

• Formulating recommendations for integrating diagnostic methods into multi-agent systems and empirically evaluating their effectiveness in a prototype.

• What expectations, requirements, and reservations do students have regarding automated profiling agents in learning environments?

• Which methods, algorithms, and data sources are discussed in the literature for ZPD estimation and adaptive scaffolding?

• How can different approaches be systematically classified and compared in terms of quality, fairness, and privacy?

• How can a profiling agent be optimally embedded into the overall system, and how can its utility be empirically demonstrated?

• Qualitative interviews or online surveys with students in the assessment year (data collection & analysis)

• Systematic literature review and/or framework analysis

• Development and evaluation of a taxonomy (e.g., following Nickerson et al.)

• Combination of approaches depending on interest

• Close scientific supervision and methodological guidance

• Access to prototypes, system data, and potentially participant pools

• Opportunity to shape the next generation of adaptive AI systems in education

• Regular meetings and support for publication efforts

Interested?
Feel free to reach out via email (diana.kozachek@unisg.ch) with a brief description of your interests and your preferred option (if applicable)!