Reproduce “ProactiveVA: Proactive Visual Analytics with LLM-Based UI Agent” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuheng Zhao, Xueli Shu, Liwen Fan, Lin Gao, Yu Zhang, Siming Chen
ProactiveVA, an LLM-based UI agent framework, offers context-aware and explainable assistance to users during visual analytics, enhancing analytical performance. It detects user difficulties through behavioral and semantic cues, providing targeted help that improves task accuracy from 0.78 to 0.95 and doubles the breadth and depth of insights.
Key insights
- Users require proactive help in specific scenarios beyond just prolonged `thinkTime`, including unfamiliarity with system functions, lack of data understanding, and failure to complete tasks accurately.
- Effective proactive assistance must be categorized and context-specific, offering onboarding tips, guidance for exploration, and reminders for errors or omissions.
- Transparency, controllability, and non-intrusiveness in the user interface are crucial for fostering user trust and enabling effective human-AI collaboration in visual analytics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.