Interaction-driven User Interface Personalisation for Mobile News Systems
User interfaces of mobile apps offer personalised experience primarily through manual customisation rather than spontaneous adaptation. This thesis investigates methods for adaptive user interfaces in the context of future mobile news apps that are expected to systematically monitor users' news access patterns and adapt their interface and interaction in response. Although mobile news services are now able to recommend news that a user would be likely to read, there has not been equivalent progress in personalising the way that news content is accessed and read. This thesis addresses key issues for the development of adaptive user interfaces in the mobile environment and contributes to the existing literature of adaptive user interfaces, user modelling, and personalisation in the domain of news in four ways. First, using survey methods it explores differences in how people consume and read news content on mobile news apps and it defines a News Reader Typology that characterises the individual news consumer. Second, it develops a method for monitoring news reading patterns through a deployed news app, namely Habito News, and it proposes a framework for modelling users by analysing those patterns; machine learning algorithms are exploited selectively in the analysis. Third, it explores the design space of personalised user interfaces and interactions that would be tailored to the needs and preferences of individual news readers. Finally, it demonstrates the effectiveness of automatic adaptation through Habito News, the prototype mobile news app that was developed, which systematically monitors users' news reading interaction behaviour and automatically adjusts its interface in response to their news reading characteristics. The results indicate the feasibility of user interface personalisation and help shape the future of automatically changing user interfaces by systematic monitoring, profiling and adapting the interface and interaction.