Evaluating intent extraction from GUIs demands accurate, fine-grained metrics. This paper introduces Bi-Fact, a novel method that decomposes intents into atomic facts and performs bidirectional comparisons to assess precision and recall. Experiments demonstrate Bi-Fact's superior correlation with human judgments compared to existing metrics, establishing a more robust evaluation framework for UI-driven intent understanding.
@article{arxiv.2502.13149,
title = {Bi-Fact: A Bidirectional Factorization-based Evaluation of Intent Extraction from UI Trajectories},
author = {Sapir Caduri and Anatoly Efros and Noam Kahlon and Danielle Cohen and Yoni Halpern and Ido Dagan},
journal= {arXiv preprint arXiv:2502.13149},
year = {2025}
}