Answers are grounded in this paper and cite their sources.
Lost architectural heritage survives not as fabric, images, or measured drawings, but as sparse classical text. We present a knowledge-constrained reasoning framework that converts such records into auditable Liao Pagoda Structural Language (LPSL) encodings. The framework combines a 3739-node Liao Pagoda Knowledge Graph (LP-KG), LPSL, case-based prototype retrieval and constrained mutation of prototype codes. Its purpose is to make an inspectable path from textual evidence to structural hypothesis, instead of a definitive original form.The framework has been validated through graph retrieval over 100 benchmark questions, LPSL stress replay, leave-one-out prototype retrieval across 68 extant pagodas and ablation under three constraint conditions. In a 100-question relational benchmark, LP-KG outperformed the lexical baseline. Top-ranked CBR outputs retained structurally relevant comparators, and progressive constraints approximately halved symbolic encoding errors. The result is a traceable candidate encoding series that distinguishes source evidence, analogy, formal grammar and conjecture before any downstream visualization.
No takes yet. Share an insight, caveat, or question.
Xu et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: