DVQA: Understanding Data Visualizations via Question Answering

TL;DR

DVQA introduces a dataset and methods for understanding bar charts through question answering, addressing the challenge of extracting numeric and semantic info from diverse visualizations.

Contribution

The paper presents DVQA, a novel dataset for bar chart question answering, and proposes baseline algorithms that outperform existing visual question answering models.

Findings

State-of-the-art VQA models perform poorly on DVQA.

Proposed baselines significantly outperform existing models on DVQA.

DVQA enables automatic extraction of information from bar charts in various domains.

Abstract

Bar charts are an effective way to convey numeric information, but today's algorithms cannot parse them. Existing methods fail when faced with even minor variations in appearance. Here, we present DVQA, a dataset that tests many aspects of bar chart understanding in a question answering framework. Unlike visual question answering (VQA), DVQA requires processing words and answers that are unique to a particular bar chart. State-of-the-art VQA algorithms perform poorly on DVQA, and we propose two strong baselines that perform considerably better. Our work will enable algorithms to automatically extract numeric and semantic information from vast quantities of bar charts found in scientific publications, Internet articles, business reports, and many other areas.