Human Perspective Based Context Acquisition, Learning and Awareness in the Design of Context Aware Systems

Context and information awareness is generally driven by the choice of available data acquisition mechanisms. In context aware system design, sensors are the primary means for data acquisition. To make such systems effective, sensors need to provide data that is accurate and reflects real time situations and events. Such data is not always readily available and may require a considerable number of in-field experiments with prototypes to gather it. In the absence of good sensor data, the situation aware models and functionality become largely ineffective. This paper proposes the use of qualitative and quantitative information gathered from logs and user research to directly generate artificial sensor data and models or complement existing sensor acquisition techniques. Data obtained via a human perspective based approach is accurate in terms of scenario requirements and user preferences, and complements sensor based data very well. This approach is similar to end-user tailoring, and preference elicitation methodologies. Furthermore, this approach allows designers to generate models that can be evaluated first before narrowing down the selection of sensors, attributes, and implementation infrastructure, thus keeping the design cost low. We present this approach with the help of a case study that involves the design of interruption aware cell phone simulations. We also propose the need for using similar human perspective based approaches in scenarios where sensor data acquisition is not accurate or feasible