حول
August 2026 HCI paper introducing a staged Generative UI framework for early design ideation. LEGOUI records prompt-derived and model-inferred decisions in a provenance-aware UI DSL, lets users accept, reject, or add decisions across layout, interaction, relation, and style stages, and renders from the evolving specification.
الملخص
Why LEGOUI Adds New Signal:
1. Inspectable intermediate state: Makes design assumptions visible before they are embedded in a generated interface
2. Decision-level control: Lets users accept, reject, or add UI-DSL items instead of repeatedly rewriting the whole prompt
3. Persistent provenance: Tracks whether each decision came from the prompt, model inference, or user revision
4. Evidence-backed workflow: Evaluates requirement extraction on 40 real-world prompts and compares the system with one-shot tools in a 15-participant study
5. Explicit trade-off: Reports stronger transparency, control, and intent alignment while acknowledging the added cost of inspecting staged decisions
1. Inspectable intermediate state: Makes design assumptions visible before they are embedded in a generated interface
2. Decision-level control: Lets users accept, reject, or add UI-DSL items instead of repeatedly rewriting the whole prompt
3. Persistent provenance: Tracks whether each decision came from the prompt, model inference, or user revision
4. Evidence-backed workflow: Evaluates requirement extraction on 40 real-world prompts and compares the system with one-shot tools in a 15-participant study
5. Explicit trade-off: Reports stronger transparency, control, and intent alignment while acknowledging the added cost of inspecting staged decisions
الوسوم
genuiui-dsldesign-timehuman-in-the-loopprovenance