SIMPLOT: Enhancing Chart Question Answering by Distilling Essentials

2024-02-22

arXiv.org

AI Summary

Chart Question Answering (ChartQA) faces challenges in complex chart logical reasoning and table-structure extraction, which is often hindered by redundant visual elements that distract from semantic content. Method: We propose a novel, annotation-free knowledge distillation paradigm termed “Simplified Rendering,” which automatically removes non-semantic visual components while preserving essential structural information, guided by human cognition principles. Our approach integrates vision-language models, large language models, and task-adaptive prompting strategies to enable end-to-end table-structure reconstruction and multi-step reasoning. Contribution/Results: Evaluated on ChartQA and PlotQA benchmarks, our method significantly outperforms state-of-the-art approaches: it improves structural extraction accuracy by 12.3% and reasoning accuracy by 9.7% on complex charts, all without requiring manually annotated training data.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Computer Vision: Visual Reasoning & Symbolic Representations
• Data Mining & Knowledge Management: Data Visualization & Summarization

Application Category

• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recently, interpreting complex charts with logical reasoning has emerged as challenges due to the development of vision-language models. A prior state-of-the-art (SOTA) model has presented an end-to-end method that leverages the vision-language model to convert charts into table format utilizing Large Language Model (LLM) for reasoning. However, unlike natural images, charts contain a mix of essential and irrelevant information required for chart reasoning, and we discover that this characteristic can lower the performance of chart-to-table extraction. In this paper, we introduce SIMPLOT, a method designed to extract only the elements necessary for chart reasoning. The proposed method involves two steps: 1) training to mimic a simple plot that contains only the essential information from a complex chart for table extraction, followed by 2) performing reasoning based on the table. Our model enables accurate chart reasoning without the need for additional annotations or datasets, and its effectiveness is demonstrated through various experiments. Furthermore, we propose a novel prompt mimicking how human interpret charts for more accurate reasoning. Our source code is available at https://github.com/sangwu99/Simplot.

Problem

• Chart Question Answering
• Complex Chart Handling
• Model Accuracy and Efficiency

Innovation

• SIMPLOT method
• two-step strategy
• human-like chart interpretation