An expert system for tax law and its educational application

The objectives of this study were to develop an expert system for an area of taxation law and to investigate issues associated with expert system development methods, explanation facilities, human -computer interface design and the use of expert systems as learning tools. An expert system giving advice on capital gains tax was developed and shown to be acceptable for use outside the development environment. Reported empirical research concerning operational expert systems is sparse and the research in this study was of an exploratory nature. System development was an incremental process, with successive

versions of the system tested with users and then modified according to test results. The explanation facility goes beyond the usual How and Why explanations offered in expert systems and is based on an analysis of the categories of questions which occur in everyday language. Intermediate steps in calculations can he viewed in a worksheet style screen or report as an explanatory device. In the development of the expert system findings and guidelines from research into human - computer interface design for more conventional systems were followed. The approach adopted led to a user -centred system rather than variable user -modelling, the avoidance of anthromorphism, and the use of menus and a form fill-in interface for input rather than natural language. Trials showed that the system gave valid advice, useful in a professional situation, and that both the explanation facility and the human - computer interface were acceptable to users. Coding for the human interface comprised almost 50% of the whole system. Previous researchers have suggested that an expert system which does not include tutoring expertise is not a good teaching device, though there is little experimental work on the topic. Results from this study indicated that the expert tax system has value as a learning tool when supported by auxiliary printed example problems and used by adults with some prior knowledge of the subject matter