A personalized and context-aware news offer for mobile devices

For classical domains, such as movies, recommender systems have proven their usefulness. But recommending news is more challenging due to the short life span of news content and the demand for up-to-date recommendations. This paper presents a news recommendation service with a content-based algorithm that uses features of a search engine for content processing and indexing, and a collaborative filtering algorithm for serendipity. The extension towards a context-aware algorithm is made to assess the information value of context in a mobile environment through a user study. Analyzing interaction behavior and feedback of users on three recommendation approaches shows that interaction with the content is crucial input for user modeling. Context-aware recommendations using time and device type as context data outperform traditional recommendations with an accuracy gain dependent on the contextual situation. These findings demonstrate that the user experience of news services can be improved by a personalized context-aware news offer