Using Process Knowledge for Adaptive User Interfaces

Abstract

In 1989 we started to develop an architecture for a more “intelligent” human-computer interface for the process control room. Having implemented a process control system for the garbage burning plant TAMARA [1,2], a test facility operated by the Kernforschungszentrum Karlsruhe (KfK), we had experiences about the special difficulties the operators have running and optimizing the plant. One of the main experiences was, that there is a lack of online-knowledge in the user interfaces currently used, and that this knowledge must not necessarily come from sophisticated diagnose systems. We can do a lot (and it is much easier than implementing a whole diagnose system) if we integrate the a-priori-knowledge we have about the plant — knowledge which is available at the time the plant is built and which describes not all but an important part of the plants behaviour.

The user interface always has at least two faces: one for the end-user and one for the designer of the user interface. Making the design effective lead us to an object-oriented architecture. As for process control, it is always necessary to visualize data from within a distributed system, we have implemented our object system as a distributed system [2,3,4]. We use a distributed object interface to connect applications (diagnose system, process interface, user interface,…).

This paper shall focus on the use of process knowledge to build adaptive user interfaces. The main goal is to optimize the dialogue at runtime according to a given process context. We use rules to make these optimizations. Rules control the dialogue and even modify the dialogue itself at runtime.