Context-aware views for mobile users ∗ Extended Abstract Cristiana Bolchini, Carlo A. Curino, Giorgio Orsi, Elisa Quintarelli, Rosalba Rossato, Fabio A. Schreiber and Letizia Tanca Dipartimento di Elettronica e Informazione – Politecnico di Milano Piazza Leonardo da Vinci, 32—20133 Milano (Italy) <surname>@elet.polimi.it Abstract Independent, heterogeneous, distributed, sometimes tran- sient and mobile data sources produce an enormous amount of information that should be semantically inte- grated and filtered, or, as we say, tailored, based on the users’ interests and context. We propose to exploit knowl- edge about the user, the adopted device, and the environ- ment - altogether called context - to the end of informa- tion tailoring. This paper presents the Context Dimension Tree, a context model which is the basis for solving the information tailoring problem, along with its role in the framework of the Context-ADDICT architecture. 1 Introduction Today we are living an epochal change, whereby the advent of the internet and the development of the commu- nication technologies have completely modified the focus of information retrieval, from the struggle for finding information and organizing it to that of appropriately reducing the enormous stream of available data. While the traditional problems typical of the data integration field are far from being solved, new challenges have also to be faced: integration of data sources which are not known in advance, automatic semantic extraction, data filtering. ∗ This research is partially supported by the Italian MIUR projects: ARTDECO (FIRB), and ESTEEM (PRIN). Mobility is, at the same time, becoming crucial for peo- ple, emphasizing old challenges while bringing to the sur- face new ones. The Context-ADDICT research addresses the above mentioned challenges, with particular emphasis on the no- tion of context: indeed, database design, especially for mobile applications, must model two different realms: the reality of interest, which is captured by the information domain model, and the user/device context. Classical data models, at a conceptual or at a logical level, are perfectly suited to represent the former, while context modeling has different demands and needs appropriate consideration. “Context” is a rather general concept and, although commonly accepted and seemingly clear, has been inter- preted in many different ways in various fields of research such as psychology, philosophy and computer science. Many proposals have been presented in these last years [1, 2, 7, 16, 21], and can be grouped in different families (some of them belonging to more than one family), based on their main focus: (i) Content presentation adaptation (focus on presentation and channel) [7, 10, 13, 15, 18] (ii) Location and environment (focus on space and situation) [10, 12, 13, 17, 18, 20] (iii) User Activity modeling (focus on what the user is doing) [14, 17, 19, 20], (iv) Context Agreement and Sharing (focus on a collectively built con- text) [9, 16] (v) Tailoring problem (focus on filtering data, services or application functionalities) [4, 22]. Although a lot of work has been done, the representation and man- agement of the context can hardly be considered to be an assessed issue. Our context model, called Context Dimension Tree, 1