Intent alignment datasets in the field of product innovation design
Published: 27 July 2026| Version 1 | DOI: 10.17632/fdthvzc222.1
Contributor:
Description
A dataset of 100 product design cases for evaluating Human-AI intent alignment in generative design. It contains fuzzy initial prompts, explicit hard constraints (product type, materials, style), and expert-annotated ground-truth design descriptions. Sourced from Behance and AI-synthesized designs, it serves as a testbed for LLM-based conceptual design, constraint satisfaction, and multi-turn interaction evaluation.
Files
Institutions
- Sichuan UniversitySichuan, Chengdu
Categories
Artificial Intelligence, Computer-Aided Design