| Interactions with the pervasive intelligence environment are increasingly becoming complex due to the presence of myriad sensors, artifacts and emerging services. Therefore, users need support to interact, control, and manage these services in their environment in an effective manner. One way to deal with this situation is to automatically identify their surrounding context and provide them context-aware services in order to reduce their cognitive load. Providing automatic support to users is known as implicit interaction, which minimizes the need for explicit command or interaction. However, as the sensor-based context identification often provides imprecise information, the automation based on such information may result in wrong actions or over automation by the intelligent environment. Therefore, the selection of inappropriate services while providing automatic support in the environment brings distrust and dissatisfaction to the people. In this thesis, we provide a framework for human-environment interactions that considers quality of context information to dynamically change the level of implicit interaction performed by the environment. Our goal is to provide a quality of context aware adaptive interaction mechanism in order to minimize the distrust and dissatisfaction of users on context-adaptive automated environment. Toward this, we first provide a model to dynamically assess the quality of context information. Based on such quality, our framework adjusts the level of output automation that the environment performs. Finally, according to the adjusted level of interaction, a set of services are either suggested or invoked for the users based on the gain in those services and their associated cost. Our experiments show that dynamically adjusting the level of interaction based on quality of context information increases the satisfaction and trust of the people in the implicit actions performed by the intelligent environment. |
