Four KG-XAI integration modes enhance defensibility in urban mining

This paper proposes a complementarity-theoretic interpretation for integrating Knowledge Graphs (KG) and Explainable AI (XAI) to support pre-demolition assessments in urban mining. The authors argue that the value of this process lies not just in prediction accuracy, but in the defensibility of decisions through legibility, plausibility, sourcing, and contestability.

The study identifies four consolidated KG-XAI integration modes: Lifting, Constraining, Typing, and Revising. Each mode is defined as a typed operation over XAI artefacts and knowledge-graph substrate structures to unlock distinct properties of defensibility. A fire-door example from the urban-mining process illustrates these modes using the W3C Linked Building Data stack.

These integration modes contribute to creating the specific regulatory artefacts required by pre-demolition assessment, addressing the structural under-specification found in existing literature.