GC

- CG is a mobile application that builds a behavioural model of the acceptance of notifications by a user.

- It ensures that the Gmail notifications are delivered at the right time. It stops a notification to trigger an alert, if the notification is predicted not to be accepted by the user at the time of delivery.

- All the filtered notifications (i.e., notifications that are stopped to trigger an alert) are shown to the user as a list of each day. The users are asked to mark the wrongly filtered notifications from this list to help us in evaluating the model.

Data collection:

- On receiving a notification, GC samples my contextual information such as location, WiFi connectivity, surrounding Bluetooth traces, current physical activity, application usage and phone status, and maps this information with the notification details such as arrival time, response time, sender app name and package. [Note: This information is used to train the model locally and never transmitted to the server.]

- We collect only the data (such as number of accepted and declined Gmail notifications, and the number of correctly and incorrectly filtered notifications) that defines the accuracy of the predictor. This information is securely transmitted and stored on the server.

- No content of the emails is used and stored by the system, i.e., no additional information would be collected by the application.