The XAI system for answer set programming xASP2.

This article focuses on the development of xASP2, an explainable artificial intelligence (XAI) system for Answer Set Programming (ASP) that generates minimal assumption sets, explanation sequences, and directed acyclic graphs (DAGs) to justify the presence or absence of atoms in answer sets. The system supports advanced ASP language features including aggregates and constraints, and produces interactive, shareable visualizations of explanations via the xASP navigator web application. The authors formalize the notion of explanations based on well-founded derivations and minimal assumptions, prove the existence of such explanations, and implement a meta-programming approach leveraging ASP solvers to compute them efficiently. Empirical evaluations demonstrate xASP2's scalability and applicability to commercial-grade ASP programs, Latin Square puzzles, and explainable planning scenarios such as the Blocksworld problem. The article also situates xASP2 within the landscape of existing ASP debugging and XAI tools, highlighting its unique combination of acyclic explanations, support for false atoms, and linguistic extensions.