Integrating Neural Networks and Semi-Markov Processes for Automated Knowledge Acquisition: An Application to Real-time Scheduling*

Recently, artificial neural networks (ANN) have gained attention as a promising modeling tool for building intelligent systems. A number of applications have been reported in areas varying from pattern recognition to bankruptcy prediction. In this paper, we present a creative methodology that integrates computer simulation, semi-Markov optimization, and ANN techniques for automated knowledge acquisition in real-time scheduling. The integrated approach focuses on the synergy between operations research and ANN in eliciting human knowledge, filtering inconsistent data, and building competent models capable of performing at the expert level. The new approach includes three main components. First, computer simulation is used to collect expert decisions. This step allows expert knowledge to be obtained in a non-intrusive way and minimizes the difficulties involved in interviewing experts, constructing repertory grids, or using other similar structures required for manual knowledge acquisition. The data collected from computer simulation are then optimized using a semi-Markov decision model to remove data redundancies, inconsistencies, and errors. Finally, the optimized data are used to build ANN-based expert systems. The integrated approach is evaluated by comparing it with the human expert and using ANN alone in the domain of real-time scheduling. The results indicate that ANN-based systems perform worse than human experts from whom the data were collected, but the integrated approach outperforms human experts and ANN models alone.     

Automated Discovery in Managerial Problem Formulation: Formation of Causal Hypotheses for Cognitive Mapping

Development of knowledge acquisition techniques known as automated discovery systems has occurred in deep and narrow domains of knowledge. Automated discovery is the generation of new knowledge by a computer system on its own, without the help of another knowledge source. This paper describes research and validation of an automated discovery system for a wide and shallow domain—business management. The system continues recent advances in expert systems research which have enhanced cognitive mapping, a problem formulation tool. The system perceives the behavior of distal variables in the environment through probabilistic cues-to-causality, and generates previously unknown hypotheses by aggregating the probabilities into a single criterion of causal relatedness. The system is validated against the source code of a simulated managerial environment, and causal relationships posited by decision makers experienced in the play of the gaming simulator.     

The Development Manager's Advisory System: A Knowledge-Based DSS Tool for Project Assessment*

This research investigates whether the knowledge-based decision support system (KBDSS) paradigm provides the necessary supporting structure and developmental framework for product development evaluation. To address the research questions posed in this study, it is necessary to develop and implement KBDSS's at specific decision points along the product development cycle. This paper describes the design, development, and implementation of a KBDSS to support a product development manager's decision concerning full-scale development of a new product. From the systems design perspective, this paper addresses the integration and innovative use of a variety of techniques for knowledge acquisition, modeling, and processing. The approach utilized obtains the benefits of normative modeling as well as the flexibility and developmental advantages of knowledge-based systems. Since its implementation, the system has been successfully used by a development manager to support his recommendation for an ongoing project. His complete satisfaction with this system served as the impetus for the design and development of a multi-expert system which was implemented at the strategic level.     

The Use of Stochastic Simulation in Knowledge-Based Systems

Knowledge-based systems support the decision-making process with the help of domain specific knowledge bases. The knowledge bases almost always have uncertainty associated with them. A variety of approaches have been proposed in the artificial intelligence (AI) literature for the construction of and reasoning with uncertain knowledge bases. Building on this stream of research, we focus on how stochastic simulation can be used to construct and reason with knowledge bases that have uncertainties. An advantage of the simulation methodology is that it may not have to make many of the assumptions made by other approaches. It also allows the designer of the knowledge-based system to control the methodology based on accuracy and time requirements. The simulation approach to knowledge base construction is a modified version of the concept induction procedure used in AI. However, it incorporates, as does simulation modeling, statistical tests to identify the best rule that describes the relationship among the variables. We show that when simulation is used to reason with uncertain knowledge bases, under certain conditions, the number of simulation trials needed to achieve a given level of accuracy is independent of the characteristics, such as the size, of the knowledge base. Empirical results obtained from an experiment confirm our theoretical results and provide evidence that simulation methodology is practical for real life knowledge-based systems.     

Automating Knowledge Acquisition and Refinement for Decision Support: A Connectionist Inductive Inference Model*

An important application of expert systems technology is to provide support for nonstructured decision making. Usually, nonstructured decision making is characterized by heavy reliance on heuristic knowledge, which is very difficult to articulate or document, and therefore traditional knowledge acquisition approaches are not very successful. The quality and effectiveness of an expert system supporting unstructured decision making is affected when traditional knowledge acquisition approaches are used. To alleviate this problem a model is proposed that combines inductive inference and neural network computing, and an example is presented that illustrates the potential of this model in unstructured decision support.     

Real Estate Taxation and Commercial Mortgage Underwriting

Recent disclosures about problem commercial real estate loans have exposed the underwriting process to intense scrutiny. This study focuses on mortgage loan underwriters of life insurance companies. After a review of the tax changes that affected commercial real estate from 1969 through 1988, the study tests how loan underwriters reacted to changes in tax benefits. To overcome the interdependent effects of interest rates and capitalization rates, a variation of the Black-Scholes pricing model is used to test the impact of changes in tax benefits. The results indicate that the underwriters do not fully incorporate the value of tax benefits in the underwriting decision. During the period of the largest tax benefits, 1982 to 1986, underwriters became more conservative and increased their equity requirements.     

Evaluating and Refining Expert Critiquing Systems: A Methodology*

Expert critiquing systems are a rapidly growing class of intelligent decision support systems that apply artificial intelligence techniques to criticize a user's proposed solution to a problem. Critic programs now exist in the medical, engineering, programming, knowledge acquisition, word processing, and other domains. Critic refinement is a nontrivial activity that, even when done well, consumes a sizable fraction of the complete effort to build and deploy a critiquing system. To ease that effort, it is important to adopt a rigorous approach that allows one to reproducibly measure the degree of success of the current critic version and to predict which refinements will improve the critic further. The current article presents one such approach with actual case studies that illustrate its usage and elaborate selected aspects of the refinement process.     

Moral Duties in Business and Their Societal Impacts: The Case of the Subprime Lending Mess

A worldwide recession of unusual breadth and depth is causing harm to many individuals and businesses. While the causes are complex, the recession got its start with failures in the subprime mortgage market. The process of creating such loans is complex and often includes securitization after loans are approved. The reasons for the subprime lending mess can be at least partly traced to decisions made by individual mortgage brokers and lending officers. These individuals had moral or ethical responsibilities because of their positions. An analysis of the rights and duties approach to ethics shows that some of these individuals failed in their moral duties and that their failure is partly to blame for the resulting harmful consequences.     

Determining Differences in Expert Judgment: Implications for Knowledge Acquisition and Validation*

In knowledge acquisition, it is often desirable to aggregate the judgments of multiple experts into a single system. In some cases this takes the form of averaging the judgments of those experts. In these situations it is desirable to determine if the experts have different views of the world before their individual judgments are aggregated. In validation, multiple experts often are employed to compare the performance of expert systems and other human actors. Often those judgments are then averaged to establish performance quality of the expert system. An important part of the comparison process should be determining if the experts have a similar view of the world. If the experts do not have similar views, their evaluations of performance may differ, resulting in a meaningless average performance measure. Alternatively, if all the validating experts do have similar views of the world then the validation process may result in paradigm myopia.     

DECISION SUPPORT SYSTEMS FOR ILL-STRUCTURED PROBLEMS: AN EMPIRICAL STUDY

Decision support systems (DSSs) are more complex than most other traditional decision-aid systems. For what types of problems are they more effective, and what design characteristics make them more effective? The laboratory experiment reported here examined the effect of three design characteristics of these systems in the context of decision makers faced with ill-structured problems. The characteristics were presence or absence of decision-aid heuristics, degree of interaction between the user and the system, and whether or not the system was computerized. The dependent variables were (1) quality of user performance, (2) user productivity of ideas, (3) user confidence in the quality of his/her performance, (4) user satisfaction with the decision aid or support system, (5) changes in user attitude toward the problem addressed, and (6) changes in user attitude toward computers. Use of heuristics and increased interaction had positive effects on decision quality, user productivity, and attitude toward computers; they had negative effects on user confidence, satisfaction, and attitude toward the problem addressed. Whether or not the system was computerized did not have a significant effect on any dependent variable. The findings concerning negative effects, in particular, suggest the need for research on the design of heuristics for addressing ill-structured problems—heuristics that will deliver the positive but not the negative effects observed in this study. The findings also suggest the need for research on how to benefit from computers in the context of solving ill-structured problems.     

A Comprehensive Model for Managing Credit Risk on Home Mortgage Portfolios*

Managing credit risk in financial institutions requires the ability to forecast aggregate losses on existing loans, predict the length of time that loans will be on the books before prepayment or default, analyze the expected performance of particular segments in the existing portfolio, and project payment patterns of new loans. Described in this paper are tools created for these functions in a large California financial institution. A forecasting model with Markovian structure and nonstationary transition probabilities is used to model the life of a mortgage. Logistic and regression models are used to estimate severity of losses. These models are integrated into a system that allows analysts and managers to depict the expected performance of individual loans and portfolio segments under different economic scenarios. With this information, analysts and managers can establish appropriate loss reserves, suggest pricing differentials to compensate for risk, and make strategic lending decisions.     

P2P网贷利率存在波动溢出吗？——基于时-频域溢出指数的实证研究

本文基于小波视角,采用多分形波动率,构建时-频域溢出指数刻画市场间复杂性,借以研究网贷市场在不同时间尺度上的波动溢出问题,并利用行为金融学和长尾理论解释异常现象,结果表明：（1）网贷市场与其他市场均存在双向溢出,且随着时间尺度从短期向长期推移,溢出强度不断上升,其中网贷市场与Shibor溢出效应最为显著。（2）总体上网贷市场主要是处于被动接受地位,但在部分短期尺度上能够主导跨市场传染过程。（3）网贷市场与其他市场间显著溢出主要集中在网贷问题频发时期。（4）在股市和债市低迷时期,网贷市场能够主导彼此间的溢出。稳健性检验佐证了以上结论。     

An Application of the AXIS Solution Framework to Multiple Objective Aggregate Production Planning

This paper provides data on the first application of a prototype of the AXIS solution framework. AXIS (algorithms combined with knowledge systems in an interactive sequence) is a framework for interactively combining structured algorithms that seek a best solution with knowledge-based expert systems that seek expert heuristic solutions. This paper tests the framework using an interactive multiple objective integer programming algorithm combined with heuristics taken from the domain of aggregate production planning. The results indicate the AXIS framework can be successful in generating high quality solutions, in vastly reduced solution times compared to the structured algorithms, at much lower costs compared to the expert heuristics working alone.     

Expert knowledge,expert systems and commercial interests

Experience to date in building expert systems has brought a general realization of the complexity of the effort required for producing systems capable of operating on ‘real-world’ problems. The most successful knowledge-based/expert systems built to date, e.g. DENDRAL, MYCIN, PROSPECTOR and R1 have demonstrated that a considerable investment in time and dedication on the part of systems designers and experts is required to create a fully operational system. The great majority of development projects attempting to harness this new programming technology do not extend beyond creation of a prototype system largely due to lack of necessary resources, i.e. time, money and know-how. The purpose of this paper is to discuss some of the more important issues associated with development of expert systems and to provide an overview of the commercial and industrial efforts of putting expert systems to work. This paper should be of interest to those who have gained their knowledge of expert systems from the reading of non-specialist publications and thus may have been exposed to somewhat over enthusiastic accounts of the subject.     

Mixed integer (0–1) fractional programming for decision support in paper production industry

This paper presents an effective and efficient method for solving a special class of mixed integer fractional programming (FP) problems. We take a classical reformulation approach for continuous FP as a starting point and extend it for solving a more general class of mixed integer (0–1) fractional programming problems.To stress the practical relevance of the research we focus on a real-life application in paper production industry. The constantly advancing physical knowledge of large scale pulp and paper production did have a substantial impact on an existing DSS in which mixed integer (0–1) fractional programming is introduced. We show that the motivation to solve a real-life fractional programming problem can provide the basis for a new approach in a new context that has an added value of its own, even outside the given application area. We describe the main characteristics of the DSS, the necessity to develop a non-iterative solution procedure and demonstrate both the effectiveness and efficiency of the proposed approach from practical data sets.     

A Computerized Approach to Decision Process Tracing for Decision Support System Design*

We propose the use of computerized process tracing (CPT) tools as an appropriate approach for monitoring the information acquisition and evaluation phase of specific decision processes. CPT tools are unobtrusive and seem particularly relevant for evaluating certain decision tasks that may be supported by decision support systems (DSS). CPT tools can be an important component of DSS development. An information systems research taxonomy developed by previous researchers [29] [36] is used to position research work involving the methodology of CPT. Using a critique suggested by Libby [28], CPT tools are evaluated and compared to alternative process tracing tools. A brief empirical example using CPT is provided, and future uses relative to DSS are suggested. The appendix includes an example of a specific CPT tool.     

TWO DESIGN PRINCIPLES FOR KNOWLEDGE-BASED SYSTEMS*

This paper discusses two principles that have become increasingly important in the design of knowledge-based systems: domain-specific knowledge used to support opportunistic reasoning and hierarchical organization structure used to control and coordinate problem-solving activity. We propose a design framework that embodies these two principles and describe how this framework has been used to develop a knowledge-based job-shop scheduling system. This system, called OPIS 0, has undergone limited testing in an experimental environment modeled after an actual job shop. Its performance has been very good compared to ISIS and to the more traditional approach of constructing a schedule by dispatching jobs using the COVERT priority rule. The resulting design also shows potential for use in a decision support role.     

DESIGNING INSPECTION STRATEGIES FOR UNCERTAIN ENVIRONMENTS

This paper considers the problems of designing inspection strategies for production systems terms in the presence of environmental uncertainty. The framework for determining information priorities to support inspection system planning is presented in the contaxt of a generic production system that encompasses the characteristics of many real-world serial production systems. The impact on the design decision of five key variables is considered: quality of producton processes, quality of inspection procedures, penalty cost for undetected defective units, relative cost of improving process vs. inspection, and shape of the cost functions for process and inspection enhancement. The framework for analysis involves varying factors over two or three orders of magnitude to determine optimal inspection strategies across a wide range of environments These results are used to compare design decisions made in the presence of environmental uncertainty using expected-opportunity-cost and minimization-of-maximum-opportunity-cost approaches. Design strategies are identified for situations ranging from complete lack of knowledge about the environment through increasing levels of information. Information-gathering priorities are established, and the impact on the design decision of this additional information is assayed.     

MULTIPLE-CRITERIA DECISION MAKING INCLUDING QUALITATIVE FACTORS: THE POST-MODEL ANALYSIS APPROACH

As a method of solving multiple-criteria decision making problems with a single quantitative objective and multiple qualitative objectives, the post-model analysis (PMA) approach is proposed. The essence of PMA is to support the trade-offs between a quantitative objective and multiple qualitative objectives so that the decision maker can find a perceived most preferred nondominated solution. To this end, the optimal solution of a quantitative model is found first, without regard for qualitative factors. The solution is then evaluated in terms of qualitative objectives. When the initial quantitatively optimal solution is adjusted to allow improvement of qualitative goals, opportunity costs of achieving qualitative goals are incurred. In this process, an expert system and/or graphical display can be used. PMA therefore provides a way to incorporate quantitative models into knowledge-based expert systems.