GitGut.ai is a voice companion application for the game League of Legends. Through real-time game analysis, it is able to amplify users' moments of happiness and support them during failure, through entertaining and educative voice comments and advices. In this field, several enhancements through machine learning are possible. In this thesis I focus on three: Trigger Reaction, analysing how the user reacts to the advices, Team Fight Detection, building a system for fight identification and In-Game Feedback, which enhance the application's logic through positive and negative feedback inputs to the concrete advices. Even though the input data is limited, I show that through careful data preprocessing and modelling we can enhance the application's logic.     «

GitGut.ai is a voice companion application for the game League of Legends. Through real-time game analysis, it is able to amplify users' moments of happiness and support them during failure, through entertaining and educative voice comments and advices. In this field, several enhancements through machine learning are possible. In this thesis I focus on three: Trigger Reaction, analysing how the user reacts to the advices, Team Fight Detection, building a system for fight identification and In-Ga...     »