Pergamon Expert SystemsWith Applications, Vol. 7, No. 2, pp. 175-183, 1994 Copyright © 1994 ElsevierScience Lid Printed in the USA. All rights reserved 0957-4174/94 $6.00 + .00 Validation of an Accounting Expert System for Business Combinations R. STEVE McDUFFIE Department of Accounting, Southwest Missouri State University, Springfield, MO L. MURPHY SMITH Department of Accounting, TexasA&MUniversity, College Station,TX STEVEN M. FLORY Department of Accounting, University of Southwestern Louisiana, Lafayette, LA Abstract--This article presents the five stages involved in the construction of a knowledge-based system using VP-EXPERT, with emphasis on the final step, knowledge validation. The knowledge- based system, called PURPOOL, enables the user to determine the proper accounting treatment for a business combination. PURPOOL can be used effectively by educators as a pedagogical device to teach their students how to account for business combinations. PURPOOL also can be used by accounting practitioners in training inexperienced staff accountants or in determining the proper accounting treatment of actual problems encountered in practice. PURPOOL is one of thefirst GAAP- based expert systems reported in the accounting literature. A GAAP-based system relies upon the current authoritative accounting literature (Financial Accounting Standards Board Statements, Ac- counting Principles Board Opinions, etc.)for the knowledge base rules. PURPOOL captures the content of APB Opinion No. 16, Business Combinations in its knowledge base. The validation of PURPOOL utilized a framework consisting of nine elements. This process included interaction with a human expert, review of the accounting literature pertaining to business combinations, development of test cases, consultations with audit seniors from a regional accounting firm, and consultations with accounting information systems and advanced accountingprofessors. As a result of using theframework, a thorough validation of the system was accomplished, and that process is described in this article. 1. INTRODUCTION IN 1987 THE AMERICAN INSTITUTE OF CERTIFIED PUBLIC ACCOUNTANTS (AICPA) published An Intro- duction to Artificial Intelligence and Expert Systems. One section of the report presented applications suit- able for future expert system development. The appli- cations included leases, foreign currency translation, acquisitions, pensions, and income taxes. Expert sys- tems are being used and developed for accounting, au- diting, and tax practice (Brown, 1991). The type of expert system development described Requests for reprints should be sent to R. Steve McDutfie, Department of Accounting, Southwest Missouri State University, Springfield, MO 65804-0094. 175 in the AICPA report departs from the typical account- ing expert systems developed thus far. The traditional accounting expert system chiefly depends upon a hu- man expert in the construction of the knowledge base of the expert system. The new type of expert system presented in the special report chiefly depends upon generally accepted accounting principles (GAAP) to provide information for the development of the rules contained in the knowledge base. These GAAP-based expert systems still rely on the human expert for system validation, but primarily rely upon the current au- thoritative accounting literature for the information to be included in the system's knowledge base. At least one such expert system has been reported in the literature (Boer & Livnat, 1990). The researchers developed an expert system to solve lease problems