TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence Reasoning

Key points

• Real-world documents distribute evidence across text, tables, figures, and captions within complex page layouts.1
• Answering complex questions over such documents therefore requires more than retrieving relevant passages: systems must recover the evidence topology that connects heterogeneous evidence units.1
• We evaluate text-only GraphRAG, page-level visual retrieval, and multimodal GraphRAG systems using retrieval, generation, and topology-aware reasoning metrics.1
• These findings motivate GraphRAG systems that move beyond text-derived entity relation graphs to explicitly model document layouts, cross-modal evidence alignment, and the reasoning roles of evidence units.1

Sources (1)

• [1]TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence Reasoning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 03:15 AM

  We introduce TOPOGRAPHRAG-BENCH, a layout-grounded benchmark for multimodal evidence reasoning in GraphRAG, comprising 2,024 questions over 201 long, visually rich documents.

  Real-world documents distribute evidence across text, tables, figures, and captions within complex page layouts.

Extractive summary: sentences quoted from the sources.