NDSS 2025 – UI-CTX: Understanding UI Behaviors With Code Contexts For Mobile Applications

This paper introduces UI-CTX, a new method for understanding the intentions of UI widgets in mobile applications. Existing methods struggle to accurately capture these intentions, often focusing on superficial features or using complex, unreliable data. UI-CTX addresses this by creating a UI Handler Graph (UHG) for each widget. This UHG incorporates relevant code context while ignoring extraneous information. The system then summarizes these graphs, considering both structural and semantic aspects to model the widget's core intentions. Extensive experiments were conducted on a large dataset of UI widget behaviors like login and search. Results demonstrate UI-CTX's effectiveness in representing UI widget intentions. It significantly surpasses current solutions in modeling these behaviors. Specifically, UI-CTX achieved a 95.2% higher F1-score than permission sets and an 8.2% higher F1-score than call sequences for classifying UI widget intentions. Furthermore, by precisely identifying widget code contexts, UI-CTX improved widget intention clustering performance by 3.6 times. The Network and Distributed System Security (NDSS) Symposium has published this content.