Structuring and Assessing Knowledge with Knowledge Space Theory

Students in the same classroom often differ substantially in what they have already mastered and what they are ready to learn next. Imagine a mathematics teacher introducing linear functions: some students can already identify the slope from a graph but struggle with determining the y-intercept from an equation, while others may not have mastered either skill yet. How can a digital learning system identify each student's specific pattern of mastered and unmastered skills and recommend appropriate next steps? Knowledge Space Theory (KST; Doignon & Falmagne, 1985) provides a mathematical framework for exactly this purpose. In KST, a domain is modeled as a set of exercise types, each capturing a specific cognitive demand. These exercise types are then organized into a knowledge structure by specifying prerequisite relations among them — that is, the dependencies that constrain which exercise types typically need to be mastered before others. Based on a student's responses, KST can infer which skills have been mastered, which are missing, and which should be learned next. This chapter introduces the formal foundations of KST and illustrates how these structures enable individualized assessment and statistically grounded insights into learning paths.