Article Preview
TopIntroduction
A central feature of ubiquitous computing applications is their capability to automatically react on context changes as to support users in their mobility (Dey et al., 2001; Henfridsson & Lindgren, 2005). Such context-awareness relies on models of specific use contexts, providing computational resources intended to facilitate user interaction in a given context. More advanced context-aware applications can also dynamically build models of their environment (Lyytinen & Yoo, 2002a). The typical objective of embedding such capabilities into computational artifacts is to support and enhance everyday use of IT over different settings.
Context-aware applications largely depend on their assumptions about user contexts. As highlighted by Dourish (2004), a common context view underlying context-aware application design is the representational one, treating context as something fairly stable consisting of a set of informational properties. This context view has proved to be useful in unambiguous contexts. Many of the seminal explorations of context-aware computing such as the Active Badge (Want et al., 1992) embed such a view. However, treating context as a set of informational properties can be constraining in more complex social settings (Chalmers, 2004; Grudin, 2001; Williams & Dourish, 2006). As highlighted by Schmidt et al. (1999), there is more to context than user location and identity. It is therefore not surprising that entire journal issues has been devoted to provide insights about how to build context-aware computing applications and architectures that cater for the typical dynamism of user contexts (see Moran, 1994: Human-Computer Interaction, Moran and Dourish 2001: Human-Computer Interaction, Schmidt et al. 2004: Computer Supported Cooperative Work). Despite the vast debate about context (see e.g., Abowd & Mynatt, 2000; Dey et al., 2001; Greenberg, 2001; Schmidt et al., 1999; Suchman, 2007), however, it has proved difficult to implement comprehensive models of context in wide-spread applications.
An important but sparsely explored aspect of context-aware computing is the role of the user as a co-creator of context. As Dourish (2004, p.22) highlights, “context isn’t just ‘there’, but is actively produced, maintained and enacted in the course of the activity at hand.” This basically means that users’ local and situated knowledge is central to the perception, as well as definition, of a given context. Acquired through previous encounters with similar situations, users’ situated knowledge is therefore part of the use setting, shaping human-computer action (Suchman, 2007). While the relevance of user experience has been highlighted in conceptual articles (Chalmers & Galani, 2004; Dourish, 2004; Greenberg, 2001), there exist few empirical studies that deal with its role in everyday use of context-aware applications.
To address this omission in the literature, this paper outlines a sequential multimethod study (Mingers, 2001) of context-aware application use for better understanding the role of user context co-creation. The study was done in the context of car navigation systems for two reasons. First, car navigation systems are a widely diffused application of context-aware computing. This enables studies of authentic use across situations characterized by different levels of situated knowledge acquired through mundane activity such as commuting. Second, car navigation systems are often highlighted as a typical example of context-aware computing (Abowd & Mynatt, 2000), involving a whole set of context indicators such as position, road classification, traffic information, and driving speed. The paper addresses the following research question: How and why does users’ situated knowledge affect everyday usage of car navigation systems?