Expert systems in spreadsheets: modeling the Wall Street user domain

Abstract

Spreadsheet models are used throughout the Wall Street user domain: one deployment requirement is that the expert system send and retrieve data to a user's spreadsheet system. The authors discuss an approach to expert system integration and deployment that is based on directly embedding expert systems within spreadsheets. This approach demonstrates expert system tool transparency: the embedded expert system is viewed by the user as another spreadsheet model. This increases user portability across applications and dramatically reduces the need for user training. They show how two types of embedded expert systems, neural network classifiers and case-based reasoning systems, may be represented as spreadsheet models. They demonstrate the embedded approach by illustrating spreadsheet expert systems for debt analysis, credit evaluation and forecasting.<>