Agent-Based User and Task Modelling for Seamless Information Access via Personal Digital Assistants

The paper describes a case study of how a combination of techniques, such as agent technologies, user, task and context modeling can be used to adapt a distributed application for mobile users with Personal Digital Assistants (PDAs) and support seamless connection and access to a centralized information system in areas of disconnection. In a domain where users have typical tasks and information needs related to these tasks it is possible, by keeping track of the users' current context (time and location through the user schedule) and a user class stereotype task model, to predict the user tasks and their information needs and to adapt both the functionality and the presentation information provided by the application. Two types of adaptation are important in a domain where there is no reliable connection: hiding from the user the disconnection and ensuring consistency of the data, which is added, deleted or modified by the user on her device during periods of disconnection. The case study’s target domain is the Saskatoon District Health Division, where homecare workers provided with iPaq PDAs equipped with CDPD cards retrieve patient information from centralized information system, make notifications for schedule changes and add new data into the information system, e.g. file reports. The prototype has been evaluated in a simulated environment with the goal to see if it provides seamless connection and access to data in conditions of disconnection.