Knowledge Acquisition, Verification, and Validation in an Expert System for Improved Traffic Safety

KLOTS (knowledge-based local traffic safety support) is a Swedish expert system that provides advice on traffic safety problems and countermeasures in urban areas. The system is briefly described and the processes of knowledge collection, verification, and validation used in its development are explained. The user defines a safety problem with input forms. The result from the system is an analysis and a list of countermeasures, each with specific comments that reflect the problem. The principle of presenting a list instead of a single solution is intended to make the user more active in the process of finding an appropriate countermeasure. In practice, KLOTS may be used for providing advice, testing solutions, and making checks. It also may be used for educational purposes. The knowledge in KLOTS was obtained from experts during interviews and is structured in the form of rules for evaluating each problem specified. Development of the system has indicated that the experts must have recent practical experience of traffic safety problems. Presenting real-world cases to the experts and asking them to explain how they would solve them has proved to be the most successful interview technique. It has been possible to achieve a consensus among experts. Extensive testing, verification, and validation are carried out before new versions of KLOTS are released. Both end user validation and knowledge verification are described. Development and widespread use of the system show both the feasibility of constructing knowledge-based systems for traffic safety and a demand for such systems.