Date
Publisher
arXiv
Interactive articles help readers engage with complex ideas through exploration, yet creating them remains costly, requiring both domain expertise and web development skills. Recent LLM-based agents can automate content creation, but naively applying them yields uncontrollable and unverifiable outputs. We present ViviDoc, a human-agent collaborative system that generates interactive educational documents from a single topic input. ViviDoc introduces a multi-agent pipeline (Planner, Executor, Evaluator) and the Document Specification (DocSpec), a human-readable intermediate representation that decomposes each interactive visualization into State, Render, Transition, and Constraint components. The DocSpec enables educators to review and refine generation plans before code is produced, bridging the gap between pedagogical intent and executable output. Expert evaluation and a user study show that ViviDoc substantially outperforms naive agentic generation and provides an intuitive editing experience. Our project homepage is available at https://vividoc-homepage.vercel.app/.
Study at a glance
- Question: Can a human-agent collaborative system generate higher-quality interactive educational documents than a single-call language model?
- Setting and data: Benchmark of 101 web-sourced interactive documents across 11 domains; expert evaluation of 20 documents; user study with three visualization or educational technology researchers.
- Method: System development, expert blind evaluation comparing two generation methods, and a user study with usability surveys and interviews.
Why it matters
Creating interactive articles remains costly and requires both domain knowledge and web development skills. Directly using automated language agents produces uncontrollable outputs and leaves educators unable to review intermediate decisions.
Key findings
- Expert evaluation showed Vividoc outperformed a naive single-call agent in content richness (4.17 versus 2.07), interaction quality (4.00 versus 2.40), and visual quality (3.73 versus 2.37).
- All three user study participants gave top usability ratings for ease of learning and ease of use, scoring both at 5.0 out of 5.
- Participants noted that the intermediate specification allowed them to control interactive behaviors without writing code, while separating text descriptions from interaction details made editing straightforward.
Summary drafted with AI from the full paper and may contain errors. See the full paper for details.
What is the application?
Who is the user?
Who age?
Why use AI?
