Activity-Driven Model for an Interactional Notion of Context

摘要

Prior research in context-awareness has largely been dominated by a positivist notion of context. While this notion of context is sufficient for well-defined and focused applications, it suffers from two main shortcomings. First, it fails to consider context as a dynamic construct that arises from a user's interactions. Second, it lacks enough consideration for the role of the human actor in context-awareness. As a result, it is inadequate for dealing with the kind of high-level activities that people naturally engage in as part of their everyday lives. This dissertation proposes an activity-driven model for an interactional notion of context that addresses these shortcomings. In this model, context is defined as a relation between activities. The model was validated using a prototype implementation running on the Google Android mobile phone emulator. Results show that not only does this model improve the computing experience of the user, it also provides unique benets that have not been available before, such as situation awareness, memory and mental aid, and an associative mode of information access. A rule-based method for discovering parent-child relationships between activities was also validated. These findings demonstrate that the activity-driven model of context warrants further research as a viable basis for context-aware mobile computing.