LLM-Based Heuristic Evaluations of High- and Low-Fidelity Prototypes (GPT-4o)

This dataset contains structured usability evaluation data for a range of user interface prototypes, assessed using Nielsen’s 10 heuristic principles. It includes:<b>Prototype_Metadata.csv</b> – Metadata for each prototype, specifying its name, fidelity level (high or low), domain, and source URL.<b>IRR_Results.csv</b> – Inter-Rater Reliability (IRR) results comparing human evaluator agreement across ten heuristics.<b>EvaluationTemplate.rtf</b> – A standardized evaluation template used by both human raters and GPT-4o to ensure consistent heuristic assessment.<b>LLM_Evaluation folder</b> – Structured evaluations conducted by GPT-4o across two phases (<i>LLM Evaluation 1</i> and <i>LLM Evaluation 2</i>), each divided into high-fidelity and low-fidelity prototypes. Each prototype file contains issue identification, severity ratings (0–4), and heuristic-based recommendations.This dataset supports research on AI-assisted usability evaluation, enabling comparison between LLM and human assessments, and analysis of prototype usability across domains and fidelity levels.