Technical integration of GSR knowledge via onboarding system prompts, knowledge graphs, retrieval-augmented generation, and a control module¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 1 study (1 qualitative), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

The authors propose concrete integration mechanisms for the identified requirements: stable characteristics such as preferred personality, student background, and learning type could be captured in a short onboarding interaction and embedded in a system prompt; dynamic information such as goals, progress, and emotional state could be updated and structured in knowledge graphs; learning materials could be accessed via retrieval-augmented generation pipelines; and interdependencies could be orchestrated by a control module such as a second LLM.

Design Implications¶

Context¶

Requirements¶

• Access to all user-related information and awareness of the physical learning environment should be optional and based on informed consent.

Constraints¶

• The authors state these mechanisms are proposals; the design requirements have yet to be implemented and empirically validated.

Target Learners¶

• university students in higher education

Target Learning Goals¶

• adaptive, grounded tutoring support for learning lecture content

Claims¶

• Participants required GSR context-knowledge of learning materials, educational strategies, and course information, with optional physical-environment perception [+W]

Related Strategies¶

• Domain-adapt RAG pipelines for mathematical education via entity recognition, notation-aware retrieval, and pedagogical re-ranking

Examples¶

-

Key Sources¶

• S. Vonschallen, D. Oberle, T. Schmiedel, and F. Eyssel. (2026). Knowledge-Based Design Requirements for Generative Social Robots in Higher Education. https://arxiv.org/abs/2602.12873