Evolutionally improved quality of intelligent systems following their users' point of view

Abstract

Authors of various intelligent expert systems concentrate their efforts on innovative elements of the developed software. But they often ignore quality aspects especially concerning software usage and users' comfort of work. The decision how to improve an expert system during its evolutional development requires an instrument to determine whether the installed software tool meets expectations of its users. One of the possible solutions, in line with agile methodologies, is to involve users in the software process. Users provide regular feedback on the considered product. In the described approach feedback is obtained in a survey by questionnaire. The guidelines are given on how to design and conduct it. Quality tree of an expert system reflects its users' point of view. Specifications formulated on the basis of the feedback allow software designers to develop improved versions of the product. The reported empirical research refers to a software system applied in civil engineering. After six iterations of its assessment and then related improvements the level of users' satisfaction from the product is currently much better than that at the beginning. © 2010 Springer-Verlag Berlin Heidelberg.