An Object Oriented Case-Based Expert System for Awarding Punishment for Serious Descipline Cases

This paper reports the application of an object-oriented case-based expert system for supporting an officer in awarding punishment for serious discipline cases in the Hong Kong Civil Service (HKCS). Over the years, there exist more than 1,500 serious discipline cases in HKCS, such as negligence of duty, conflict of interest, misbehavior, fraud and dishonesty. The determination of punishment is a very time consuming and error prone process. Furthermore, management also has to balance the interests and demoralizing effect that may arise from the decision. By building an object-oriented casebased expert system to tackle this problem, the efficiency and effectiveness of coming to a decision has been greatly improved. Basically, previous discipline cases were organized into an object hierarchy using the common features for determination of class membership. Retrieval of cases is done in two phases, in phase one, primary features are matched. If they are successfully matched, secondary features will be used with weighting factors to find out the best match. The solution adaptation is done by firing the production rules captured during the knowledge acquisition stage, the result would be an adapted solution based on the parameters from the new input case. The system was built using an object-oriented expert system shell, KAPPA-PC, and the prototype was tested by existing officers. The results illustrated that the system is a very useful aid for supporting their decision.