Compare alternative representations of knowledge components

Investigate whether alternative representations of knowledge components—including different levels of granularity, abstraction, relationships, generative representations, and AI-driven personalization—improve HelpCoach’s coverage and support flexible levels of knowledge specificity.

Background

HelpCoach uses a fixed set and granularity of predefined knowledge components. Some participants reported that these options did not cover all topics they wanted to ask about, while finer-grained or more numerous components could increase cognitive load. The paper leaves unresolved whether continuous, dynamically generated, or user-created representations would better accommodate relationships among components and differing levels of specificity.