Testing, Verification, and Validation of Expert Systems

Toward More Effective Transportation Applications of Computational Intelligence Paradigms

“…Since verification of the knowledge-based system consists of checking the completeness and consistency of the knowledge base and the quality of its inferencing mechanism, it is essentially a programming task and therefore is more amenable to traditional software engineering approaches (4,(6)(7)(8)(9)(10). This is especially true for the datadriven systems.…”

“…This is especially true for the datadriven systems. In contrast, CI system validation has been mostly ad hoc, informal, and of dubious value (4,10). In recent years, a variety of approaches to validation have been described in the literature.…”

“…How closely the conclusions agree is used as a measure of performance. Methods include simple comparisons of case study outcomes (10,24), hypothesis tests ( 15), analytical models [for example, the outputs of neural network models have been compared to those from regression models and from linear discriminant models ( 25)], and simulation models. Simulation model methods are analogous to the simple case study method, where it is not possible to use case studies, such as real-time applications like traffic-control systems.…”

A web-based information system for geoconstruction technologies and performance of stone column reinforced ground

“…Each selection queries a database column and utilizes a nested if…then statement to sort the appropriate technologies. A significant benefit of the rule-based approach is the sharing of knowledge, especially when the knowledge is not the type of knowledge typically published in scholarly publications (Spring et al 1991). The knowledge for identifying potentially applicable technologies to a set of geotechnical and project conditions was initially developed from the summaries and assessments for each of the technologies.…”

“…Only technologies that satisfy all of the queries remain as potentially applicable technologies. A significant benefit of the rule-based approach is the sharing of knowledge, especially when the knowledge is not the type of knowledge typically published in scholarly publications (Spring et al 1991). Fuzzy logic and probability theory were considered for use in the development of the interactive selection system.…”