Optimizing and automatically assessing soft skills within knowledge-graph-based adaptive learning

Determine whether and how non-epistemic competencies (soft skills), including abilities such as critical thinking and creating new value, can be optimized and feasibly measured within knowledge-graph-based adaptive learning systems by developing error-correcting and automated assessment methods for these competencies.

Background

Adaptive learning platforms typically rely on domain content models structured as knowledge graphs or knowledge spaces, which are well-suited for machine-assessable, subject-specific knowledge and skills. This architecture constrains what these systems can effectively teach and evaluate.

Modern education places increasing emphasis on general competencies or so-called 21st-century skills (e.g., critical thinking, collaboration, and creating new value). Integrating these non-epistemic competencies into knowledge-graph-driven systems and assessing them automatically pose significant difficulties, highlighting the need for methods that both optimize such competencies and provide valid automated, error-correcting assessments.