The implementation of knowledge based systems into organizations

A knowledge based system in applied physiology was developed over a period of four years for the Ministry of Defence. It was intended for use by the Applied Physiology (AP) Division of the Army Personnel Research Establishment (APRE), Farnborough, Hampshire. The system was named MAPS (Modular Applied Physiology System). The aim was both to develop MAPS, and to produce guidelines on the implementation of knowledge based systems into organizations, based on the experiences gained during the project. The main body of the document is therefore a case study, providing details of the decisions made at each stage of development. MAPS was developed on-site, using existing computer hardware and software. The design plan adopted was an initial early demonstration system, followed by an iterative process of prototype development. Particular attention was paid to the collection of user feedback, and user participation from the outset of the project. There were three systems during the development cycle, the demonstration system (MAPSl), and two full prototypes (MAPS2 and 3). The user interface evolved gradually over the three versions, the criteria being to develop an interface suitable for infrequent, and computer naive users. MAPS was evaluated before the introduction of each subsequent version. A selection of example enquiries was required to be answered using MAPS. The evaluations of MAPSl and MAPS2 were intended to serve a dual purpose, performing the additional role of a tutorial. A questionnaire containing 7 point adjective scales and open ended questions was used to collect subjective views from the user. The results are tabulated, and presented graphically. Their implications for systems design are discussed. Potential pitfalls during development were identified, in particular: assessing the level of computer knowledge possessed by the users; Keeping users up to date on a constantly expanding knowledge base; and aids to navigation through the tree structure of such knowledge. The thesis concludes by proposing a range of ten broad guidelines aimed at helping future software developers, when faced with a similar task. They are intended for a wide audience, and are appropriately general, and largely non-technical.