Knowledge Technology Applications for Knowledge Management

Abstract: We investigate how the knowledge in knowledge technology applications for knowledge management can be represented to let the user directly manage the knowledge content of the applications.In paper I we design a representation of diagnosis knowledge that allows the user to add new components and inspect the function of the device. The representation allows an integration of model based knowledge with compiled and heuristic knowledge so that the device and its function can be represented a suitable level of abstraction and let other parts be represented as non-model based knowledge.In paper II we use simplified rules for describing the time, resources, activities and amounts required in a tunnelling project and a simulation engine for estimating time and amounts consumed in the tunnelling process. The rules are designed to allow a user to change the facts and computations of the system.In paper III we present the constraint modelling language CML and show how to model a nurse scheduling problem and a train scheduling problem without programming. The idea is to preserve the problem structure of the domain, allowing constraint formulations that reflect natural language expressions familiar to the users. CML problem specifications are transformed automatically to standard constraint programs In paper IV we investigate the use of decision tables for representing requirements on staff scheduling explicitly, providing structure, decision support and overview to the user. The requirements are compiled automatically to a program that use hand-written procedures for efficient scheduling.It seems possible to let the user modify central parts of the knowledge content in the applications with these representations, by using various illustration techniques. The techniques used are object-based graphics for manipulating device components and connections in diagnosis, simplified rules for simulation of tunnelling activities, text-based query language specification of scheduling problems and finally, decision tables for constraint problems and decision support.