Supplementary Material: An Exploratory Study on Prompting Strategies for LLM-Assisted Heuristic Evaluation of Web User Interfaces

User Interfaces (UIs) — The 12 AI-generated e-commerce web interfaces used as evaluation targets in the study. Dataset — The set of Nielsen heuristic violations identified for each interface, comprising both the LLM-generated evaluations (GPT under zero-shot, few-shot, and chain-of-thought prompting) and the ground-truth annotations produced by a senior usability specialist, used as the basis for comparison. Analysis Code — A Jupyter notebook (.ipynb) implementing the full analysis pipeline, including the computation of agreement metrics (F1-score and Cohen's Kappa) between the LLM-generated evaluations and the specialist's ground truth across the three prompting strategies.