MisVisFix: An Interactive Dashboard for Detecting, Explaining, and Correcting Misleading Visualizations using Large Language Models

TL;DR

MisVisFix tackles the challenge of misleading visualizations by delivering an end-to-end interactive dashboard that detects, explains, and automatically corrects chart misinformation using dual multimodal LLMs (Claude and GPT). It achieves a high detection performance (F1 = 0.96) across 74 issue types, provides precise issue localization, and generates corrected visualizations while enabling user-driven learning and refinement. Expert evaluations validate practical usefulness in professional and educational contexts, highlighting its potential to enhance visualization literacy and data communication integrity. The work also outlines a pathway for social-media integration and future improvements, including domain customization and performance optimizations.

Abstract

Misleading visualizations pose a significant challenge to accurate data interpretation. While recent research has explored the use of Large Language Models (LLMs) for detecting such misinformation, practical tools that also support explanation and correction remain limited. We present MisVisFix, an interactive dashboard that leverages both Claude and GPT models to support the full workflow of detecting, explaining, and correcting misleading visualizations. MisVisFix correctly identifies 96% of visualization issues and addresses all 74 known visualization misinformation types, classifying them as major, minor, or potential concerns. It provides detailed explanations, actionable suggestions, and automatically generates corrected charts. An interactive chat interface allows users to ask about specific chart elements or request modifications. The dashboard adapts to newly emerging misinformation strategies through targeted user interactions. User studies with visualization experts and developers of fact-checking tools show that MisVisFix accurately identifies issues and offers useful suggestions for improvement. By transforming LLM-based detection into an accessible, interactive platform, MisVisFix advances visualization literacy and supports more trustworthy data communication.

Figures (5)

• Figure 1: Example output from the MisVisFix issue analysis interface for a specific misleading visualization
• Figure 2: Visualization generation workflow using Claude and GPT. The system analyzes uploaded charts, generates Python code for improved versions, and enables iterative refinement through interactive chat until users are satisfied with the results.
• Figure 3: Comparison of original misleading visualizations and MisVisFix-generated corrections. Top: Pie chart with too many similar-colored segments converted to sorted bar chart after detecting inappropriate chart type, indistinguishable colors, and missing units. Bottom: Line chart with dual y-axes causing data misrepresentation replaced with single-axis line charts after identifying dual axis issues, data magnitude differences, and misrepresentation problems. Both examples show conversion to clearer chart formats that reduce misinterpretation and improve data comprehension.
• Figure 4: MisVisFix learning mechanism interface demonstrating how users can add undetected issues to the system's knowledge base.
• Figure 5: Proposed 'Truthify' toggle feature for social media integration. Users can switch between the original misleading visualization and the corrected version directly in their social feed, with highlighted issues explaining specific problems.