Integrating Answer Set Programming and Item Response Theory for Adaptive Assessment in E-Learning

Adaptive assessment systems should accurately estimate learner ability while also satisfying pedagogical require- ments such as prerequisite relationships, balanced topic coverage, and personalized learning paths. This paper presents an Adaptive Quiz Framework for personalized e-learning assessment that integrates Item Response Theory (IRT) and Answer Set Programming (ASP). IRT provides a statistically grounded estimate of learner ability and guides adaptive item selection, while ASP enables declarative, constraint-based reasoning to enforce prerequisite relationships within a topic dependency graph, maintain content balance, and satisfy additional personalization constraints that are difficult to capture through purely statistical item selection strategies. The framework dynamically selects the next question according to the learners estimated ability, item difficulty, and associated learning topic. Quiz content can be sourced from existing question banks or created through a graphical authoring interface, optionally supported by a Large Language Model (LLM) to generate topic structures, author assessment items, and provide preliminary difficulty estimates. The proposed integration of adaptive modeling and declarative reasoning demonstrates how adaptive assessment can jointly address measurement accuracy and pedagogical coherence.