CoDA: Agentic Systems for Collaborative Data Visualization

2025-10-03

AI Summary

Existing automated visualization systems struggle with multi-file, complex datasets and iterative refinement requirements, often failing due to oversimplified task decomposition, inadequate error handling, and absence of quality feedback. This paper proposes a large language model–based collaborative multi-agent system that decomposes visualization generation into four synergistic modules: metadata parsing, task planning, code generation, and quality-driven self-reflection. By integrating metadata analysis and introspective mechanisms, the system overcomes context-length limitations and enables end-to-end iterative optimization. Its core contribution is a verifiable and correctable multi-agent workflow specifically designed for complex data scenarios. Experiments demonstrate that our approach achieves a 41.5% improvement in composite evaluation scores over strong baselines, significantly enhancing both success rate and output quality for visualization generation under complex conditions.

Technology Category

• Multiagent Systems: Multiagent Planning
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Deep research has revolutionized data analysis, yet data scientists still devote substantial time to manually crafting visualizations, highlighting the need for robust automation from natural language queries. However, current systems struggle with complex datasets containing multiple files and iterative refinement. Existing approaches, including simple single- or multi-agent systems, often oversimplify the task, focusing on initial query parsing while failing to robustly manage data complexity, code errors, or final visualization quality. In this paper, we reframe this challenge as a collaborative multi-agent problem. We introduce CoDA, a multi-agent system that employs specialized LLM agents for metadata analysis, task planning, code generation, and self-reflection. We formalize this pipeline, demonstrating how metadata-focused analysis bypasses token limits and quality-driven refinement ensures robustness. Extensive evaluations show CoDA achieves substantial gains in the overall score, outperforming competitive baselines by up to 41.5%. This work demonstrates that the future of visualization automation lies not in isolated code generation but in integrated, collaborative agentic workflows.

Problem

• Automating data visualization from natural language queries
• Managing complex multi-file datasets and iterative refinement
• Ensuring robustness against code errors and quality issues

Innovation

• Multi-agent system for collaborative data visualization
• Specialized LLM agents handle metadata and code
• Quality-driven refinement ensures robust automation