![]() |
|||
![]()
|
![]() |
![]() |
![]() |
3.2.5. PARMENIDE PARMENIDE was developed for the Italian Banco di Napoli. It reviews loan applications on the basis of its prediction of the future position of the company applying for the loan facility. In order to assess the working conditions, assets, etc., the customer is visited during the assessment. The management, background of the company, and its market are incorporated into the analysis. Information from external agencies is used in order to incorporate market data. While it is an example of a move toward the use of qualitative data and related heuristics, much of the analysis requires expertise on the part of the loan officer and the assessment is, to some extent, made in the absence of detailed quantitative knowledge of the sector in which the company operates. It guides the loan officer through the entire process, but requires input of generic risk-related expertise, part of which is provided centrally by the Bank's lending expert, not the loan officer dealing with the application, during the second phase of the assessment. Reference: Condensed from Butera, G., Frascari, E., and Iacona, G. (1990). Parmenide: an expert system for assessing the credit of industrial clients. International Journal of Expert Systems. Vol. 3, 1990, 73-85. 3.2.6. KABAL KABAL is a bank loan authorization expert system developed with the Norwegian Tromsø Sparebank. It performs analysis of financial statements. It also considers guarantees, market, and company management/organization. The user is guided through the appraisal but there are many points where the loan officer is required to have relatively deep expertise. Reference: Condensed from Hartvigsen, G. KABAL: a knowledge-based system for financial analysis in banking. Expert Systems for Information Management. Vol. 3, 1990, 213-31. 3.2.7. EVENT Money lending is one of the main acitivities of all banks. A bank must decide whether or not to grant a loan by judging the customer's ability to repay the loan according to lending guidelines established by the bank. This task may be repeated often for different customers, requiring select bank personnel who must repeat the same routine steps when processing the loans. Evalog, a French bank, faced the situation when processing large loans to companies and decided to investigate the application of expert systems to ease the workload. The system they developed, EvEnt, judges a company's overall credibility and performance when making its recommendation. It considers such factors as the company's financial structure, size, and management performance to evaluate the risk in lending the company funds. EvEnt decreased the cost of processing loans tenfold, helped Evalog process more loan applications, and helped minimize the bank's exposure to risk. Reference: Durkin, J.: Expert systems: a view of the field. IEEE Expert, Vol. 11 (2), 1996, 56-63. 3.3. CAPITAL INVESTMENT3.3.1. Management Advisor Management Advisor is used for decisions involving financial investment in new business opportunities. It is intended for use by executives and managers. It considers financial issues, risk, timing, competition, and the organizational impact. Among the financial factors considered during a consultation are prices, market share, costs, depreciation, and taxes. Users can customize the expert system to match their own terminology, financial assumptions, and preferences. The system comprises around 200,000 lines of code and is written in PSL, a LISP-based language. Reference: Condensed from Reitman, W. and Shim, S.J. Expert systems for evaluating business opportunities: implementing the management advisor at Krypton Chemical, Intelligent Systems in Accounting, Finance and Management. Vol. 2, 1993, 191-204. 3.3.2. AFFIN AFFIN is used by bank managers for the evaluation of industrial investment projects. It determines the feasibility and economic convenience of a proposed project. Included in the analysis are consideration of productive, plant location, and technology aspects, along with company capability (managerial and employee) and company financial factors. AFFIN performs a market analysis considering product description, market size and segmentation, and forecasted sales. It was developed in C and is backward-chaining, using both rules and semantic nets. Reference: Condensed from Durkin, J. Expert Systems Catalog of Applications. Intelligent Computer Systems Inc: Akron, OH, 1993, 16. 3.3.3. Cash Flow Profiler The system was written using the Leonardo Level III expert system shell. It is rule-based and includes links from the shell to external packages -- a DBMS and a spreadsheet. The system produces an estimate of the cash flow profile of capital projects, thereby enabling decisions on, for example, whether to proceed or not to be taken. The estimates are at three levels: best guess, lower limit, and upper limit. Data input is via a macro-driven spreadsheet. Forward-chaining is employed. The knowledge base was built using the cash flow profiles of 76 projects along with semistructured interviews with four domain experts. As more projects are assessed using the system, the actual outcomes are added to the database and the paramenters adjusted accordingly. Reference: Condensed from Lowe, J.G., Moussa, N., and Lowe, H. Cash flow management: an expert system for the construction client, International Journal of Applied Expert Systems. Vol. 1, 1993, 134-152. 3.3.4. PROJECT PROJECT is used by local authorities in France to test the financial impacts of a portfolio of projects. Its knowledge base includes information on the legal aspects of local taxation, and on operating costs associated with generic investments, such as schools. Reference: Condensed from Klein, M. and Methlie, L.B. Expert Systems: a decision support approach. Addison Wesley: Wokingham, 1990, 465-6. 3.4. PREDICTING STOCK MARKET BEHAVIORThe stock market behavior prototype system developed by Braun and Chandler (1987) using a rule-induction approach is an attempt to provide stock market analysts with a tool that combines both quantitative and categorical measures. Statistical techniques, such as discriminant analysis and regression analysis, are suitable when analyzing the quantitative measures, while techniques such as logit and probit analysis can address the categorial measures to some extent.
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |