Monte Carlo Tree Search for Table-to-Multimodal Report Generation

2026-08-04

AI Summary

Existing approaches to table-to-multimodal-report generation struggle to simultaneously ensure factual accuracy, visual quality, and narrative coherence due to their rigid linear pipelines and isolated subtask processing. This work proposes MCTS-Report, the first framework to integrate Monte Carlo Tree Search (MCTS) into this task, modeling report generation as a progressive construction process within a structured action space. The framework orchestrates large language models to collaboratively execute atomic operations—including section planning, chart generation, and insight organization—augmented by dynamic reasoning and context-aware mechanisms. By incorporating a multidimensional reward function, diversity penalties, and precondition-based pruning, the method substantially enhances output quality. Evaluated on MMRBench, a newly curated benchmark spanning six domains, MCTS-Report consistently outperforms strong baselines across structural completeness, numerical fidelity, text-image alignment, and insight novelty, achieving a comprehensive score of 77.9.

Abstract

Automatically generating professional multimodal reports comprising both textual analysis and visual charts from structured tabular data is a critical challenge in data intelligence. Existing methods suffer from fixed linear pipelines and isolated subtask processing, which hinder joint optimization of factual accuracy, visual quality, and narrative coherence. To address these issues, this paper proposes MCTS-Report, a Monte Carlo Tree Search (MCTS)-driven framework that formulates multimodal table-to-report generation as a progressive construction process over a structured search space. The core idea is to decompose report generation into atomic actions, including chapter planning, visualization task identification, chart generation, insight organization, and narrative refinement, each executed by an LLM based on dynamic reasoning conditioned on the current report state. We use an LLM to generate step-by-step reasoning and actions during MCTS, storing the reasoning trajectory in each node for context-aware, coherent report construction. To guide the search, we design a multi-dimensional reward function that jointly evaluates numerical fact consistency (via SQL), chart quality, chart-text alignment, and structural completeness, while incorporating a diversity penalty to suppress repeated charts and a precondition check to prune invalid actions. We also construct MMRBench, a comprehensive benchmark comprising real-world tables from six domains, paired with expert-refined reference report structures and verifiable key insights. Experiments on MMRBench demonstrate that MCTS-Report significantly outperforms strong baselines across structural completeness, numerical accuracy, chart-text alignment, and insight novelty, achieving a 77.9 overall score.

Problem

• multimodal report generation
• table-to-text
• data visualization
• factual consistency
• narrative coherence

Innovation

• Monte Carlo Tree Search
• Multimodal Report Generation
• Large Language Model
• Structured Table Understanding
• Reward-guided Search