FlowEval: Reference-based Evaluation of Generated User Interfaces
For developers and researchers using LLMs for UI generation, FlowEval provides an automated evaluation method that correlates with human expert judgments, reducing reliance on slow and costly human evaluation.
FlowEval is a reference-based framework that evaluates generated user interfaces by comparing navigation traces from real websites to generated analogs using similarity metrics like dynamic time warping. In a small-scale study, reference-based metrics strongly correlated with human judgments, suggesting scalable and trustworthy evaluation for UI generation.
While large language models (LLMs) and coding agents are often applied to user interface (UI) development, developers find it difficult to reliably assess their proficiency in visual and interaction design. Existing evaluations either rely on human experts, who can accurately assess usability by testing critical flows but are slow and costly, or on automated judges, which are scalable but less accurate and opaque. We present FlowEval, a reference-based framework that measures whether a generated UI supports realistic interaction flows by comparing navigation traces from real websites to traces from generated analogs using reference-based similarity metrics (e.g., dynamic time warping). In a small-scale study with expert UI evaluators, we show that reference-based metrics strongly correlate with human judgments, suggesting that they can provide scalable yet trustworthy evaluation for UI generation systems.