Abstract: The subject technology provides generating configuration files that represent user interaction scenarios on a device, the configuration files including a set of user interface (UI) elements to display and a set of interactions corresponding to user activity to perform on at least one UI element of the set of UI elements. The subject technology processes the configuration files using a relevance engine to determine a relevance score of each UI element of the set of UI elements, wherein the relevance score is based at least in part on weights assigned to features based on the user activity and respective variance values of the features, and the features include information related to a location or time. The subject technology generates a user simulation based machine learning model based at least on each relevance score of each UI element. The subject technology stores the user simulation based machine learning model on the device.

Abstract: The subject technology provides generating configuration files that represent user interaction scenarios on a device, the configuration files including a set of user interface (UI) elements to display and a set of interactions corresponding to user activity to perform on at least one UI element of the set of UI elements. The subject technology processes the configuration files using a relevance engine to determine a relevance score of each UI element of the set of UI elements, wherein the relevance score is based at least in part on weights assigned to features based on the user activity and respective variance values of the features, and the features include information related to a location or time. The subject technology generates a user simulation based machine learning model based at least on each relevance score of each UI element. The subject technology stores the user simulation based machine learning model on the device.