Context-Aware Software

With the advent of PDAs (Personal Digital Assistants), smart phones, and other forms of mobile and ubiquitous computers, our computing resources are increasingly moving off of our desktops and into our everyday lives. However, the software and user interfaces for these devices are generally very similar to that of their desktop counterparts, despite the radically different and dynamic environments that they face. We propose that to better assist their users, such devices should be able to sense, react to, and utilise, the user's current environment or context. That is, they should become context-aware. In this thesis we investigate context-awareness at three levels: user interfaces, applications, and supporting architectures/frameworks. To promote the use of context-awareness, and to aid its deployment in software, we have developed two supporting frameworks. The first is an application-oriented framework called stick-e notes. Based on an electronic version of the common Post-It Note, stick-e notes enable the attachment of any electronic resource (e.g. a text file, movie, Java program, etc.) to any type of context (e.g. location, temperature, time, etc.). The second framework we devised seeks to provide a more universal support for the capture, manipulation, and representation of context information. We call it the Context Information Service (CIS). It fills a similar role in context-aware software development as GUI libraries do in user interface development. Our applications research explored how context-awareness can be exploited in real environments with real users. In particular, we developed a suite of PDA-based context-aware tools for fieldworkers. These were used extensively by a group of ecologists in Africa to record observations of giraffe and rhinos in a remote Kenyan game reserve. These tools also provided the foundations for our HCI work, in which we developed the concept of the Minimal Attention User Interface (MAUI). The aim of the MAUI is to reduce the attention required by the user in operating a device by carefully selecting input/output modes that are harmonious to their tasks and environment. To evaluate our ideas and applications a field study was conducted in which over forty volunteers used our system for data collection activities over the course of a summer season at the Kenyan game reserve. The PDA-based tools were unanimously preferred to the paper-based alternatives, and the context-aware features were cited as particular reasons for preferring them. In summary, this thesis presents two frameworks to support context-aware software, a set of applications demonstrating how context-awareness can be utilised in the ''real world'', and a set of HCI guidelines and principles that help in creating user interfaces that fit to their context of use.