Mobile application for context inference at entertainment events

In thesis we develop an Android application for context inference at entertainment events. The application detects a user's physical activity and location, a number of people in the vicinity, and music that the user is listening to. The application uses multiple sensors available on the phone, such as microphone, accelerometer, Bluetooth, and others, and presents the inferred information in a user-friendly form. The application then allows the user to share these information with other users of the app. For the implementation of the above functionalities, our app uses Google Play Application Programming Interface for detecting location and user activity; Bluetooth for detecting a number of people in vicinity; and Dejavu Python library for detecting music. Besides the Android app, our solution encompasses a server and a database. The mobile application is used for user interaction and detection, the database stores all the data that our application needs, while the server handles communication between the application and the database. We test the developed solution among ten users who evaluated its usability and ability to correctly detect the context. The evaluation enables us to extract actionable guidelines for future improvements in mobile-based context detection for social/entertainment applications