Nowadays people are surrounded by an enormous amount of information, therefore a number of approaches have been developed to select the news content that is relevant to the user. Much less research has been done on the topic of adapting the way the content is shown to the user. To address this gap, in this thesis we devise an approach for automatic context-aware personalised mobile news display adaptation. To achieve this, we develop a full-fledged mobile newsreader adaptation that also includes context sensing and user experience querying and conduct two real-world studies with this application. The purpose of the first study is to collect data used for multilevel hierarchical modelling of the relationship between the context and the user's news reading preferences. We find that a user’s physical activity and the environmental brightness have the greatest influence when predicting the values of the preferred news display parameters. We then proceed with the construction of predictive models. However, a major weakness of such models is that they do not take into account the diversity of user’s preferences. This we address through the second study, where in the begining we focus on the personalization of existing predictive models, while later we compare the personalized and the general models. Our studies show that users have diverse needs, leading to higher prediction power of personalized models.
|
