Reading Order Inference for Complex Document Layouts

2026-07-01

AI Summary

This study addresses the challenge of accurately inferring reading order in complex historical manuscripts—such as the Glossa Ordinaria—where annotations irregularly surround the main text, confounding existing layout analysis methods. The authors propose a training-free graph-based reasoning approach that constructs a directed candidate transition graph from OCR text lines, weights edges using signals from a causal language model and BERT’s next-sentence prediction, and recovers the global reading sequence via degree-constrained path cover combined with a max-regret inference rule to avoid greedy errors. This method uniquely integrates lightweight language model signals with max-regret reasoning, achieving 95% average recovery of true successor edges on Glossa layouts (versus 50% for XY-cut) and 88% macro-edge accuracy on the OmniDocBench multi-column subset—substantially outperforming LayoutReader (25%) and XY-cut (75%)—while exhibiting mirror invariance.

Abstract

Reading order inference remains a critical bottleneck in the digitization of complex historical manuscripts, where pages contain multiple spatially interleaved reading streams, the canonical example being the Glossa Ordinaria layout, in which a central text is surrounded by commentaries that wrap around it in non-rectangular, non-convex regions. We present a training-free, graph-based framework: each OCR text line becomes a node in a directed candidate-transition graph, edges are scored by a weighted additive ensemble of two lightweight language-model signals (causal language model conditional likelihood and BERT next-sentence prediction, NSP; a third sentence-embedding signal was evaluated but did not improve reading order), and the global reading order is recovered as a degree-constrained directed path cover. To avoid the cascading "edge-theft" failures of greedy edge selection, we propose a max-regret inference rule that prioritizes commitments with high opportunity cost. We evaluate on synthetic Glossa Ordinaria grid layouts, on 23 ALTO page geometries (10 historical source pages plus mirrored and flipped variants), and on a 140-page multi-column English subset of OmniDocBench, comparing our method against the canonical recursive XY-cut (PaddleOCR PP-StructureV3) and two LayoutReader variants (layout-only and text+layout) on identical inputs. On wrap-around Glossa layouts our method recovers 95% of ground-truth successor edges on average vs. XY-cut's 50%; on the OmniDocBench multi-column subset it reaches 88% macro edge accuracy versus XY-cut's 75% and LayoutReader's 25%. The LayoutReader baselines transfer poorly due to a word-level vs. line-level granularity mismatch. We additionally verify mirror-invariance under horizontal and vertical page reflections: Our method changes by less than 1 percentage point, classical XY-cut by 2 points, and LayoutReader-T by up to 8 points.

Problem

• reading order inference
• complex document layouts
• Glossa Ordinaria
• spatially interleaved reading streams
• historical manuscripts

Innovation

• reading order inference
• graph-based framework
• max-regret inference
• language model ensemble
• degree-constrained path cover