Showing papers in "Expert Systems With Applications in 1999"

Neural networks in business: a survey of applications (1992–1998)

Abstract: During the last decade, neural networks have established themselves as a theoretically sound alternative to traditional statistical models, and a large body of research on their application to business has been produced. The comprehensive range of business and financial applications is such that a focus is required for an in-depth analysis, therefore this review addresses applications related to management, marketing and decision making. Also, given that previous reviews have dealt with earlier publications, the time span of the review is limited to the period 1992–1998. The presentation is centred on summary tables with links between them. These tables classify the studies according to their application areas, the main contributions rendered by the use of neural networks, and the alleged advantages and disadvantages of this, as well as the journal of publication. Further information on the neural network models, other statistical methods against which they have been compared, and features of the analysed data are also provided. The more controversial issues concerning real-world applications of neural networks are discussed as a part of a critical analysis. Many of the studies are shown to be first attempts to apply these new techniques to established areas of research, whereas only a few tackle real-world cases. Although still regarded as a novel methodology, neural networks are shown to have matured to the point of offering real practical benefits in many of their applications.

Case-based reasoning supported by genetic algorithms for corporate bond rating

Abstract: A critical issue in case-based reasoning (CBR) is to retrieve not just a similar past case but a usefully similar case to the problem. For this reason, the integration of domain knowledge into the case indexing and retrieving process is highly recommended in building a CBR system. However, this task is difficult to carry out as such knowledge often cannot be successfully and exhaustively captured and represented. This article utilizes a hybrid approach using genetic algorithms (GAs) to case-based retrieval process in an attempt to increase the overall classification accuracy. We propose a machine learning approach using GAs to find an optimal or near optimal weight vector for the attributes of cases in case indexing and retrieving. We apply this weight vector to the matching and ranking procedure of CBR. This GA–CBR integration reaps the benefits of both systems. The CBR technique provides analogical reasoning structures for experience-rich domains while GAs provide CBR with knowledge through machine learning. The proposed approach is demonstrated by applications to corporate bond rating.

Segmentation of the on-line shopping market using neural networks

Abstract: The characterization and analysis of on-line customers' needs and expectations, regarding the Internet as a new marketing channel, is considered a prerequisite to the realization of the expected growth of the consumer-oriented electronic commerce market. The aim of the present study is twofold: to carry out an exploratory segmentation of this market that can throw some light upon its structure, and to characterize the on-line shopping adoption process. The Self-Organizing Map (SOM), an unsupervised neural network model devised by Kohonen (Kohonen, T. (1982). Self-organized formation of topologically correct feature maps. Biological Cybernetics, 43(1), 59–69; Kohonen, T., (1995). Self-organizing maps. Berlin: Springer) will be used as part of a tandem approach to segmentation, which involves the factor analysis of the observable variables in the data to be analyzed, prior to clustering. The SOM is shown to be a powerful data visualization tool, able to assist the data analysis, providing supervised methods with useful explanatory capabilities. It is also applied, in a completely unsupervised mode, to discover the clusters or segments that naturally occur in the data. The SOM is proposed as a flexible clustering model able to accommodate both Finer Segmentation and Normative Segmentation approaches. Within the latter, a cluster-partition is proposed and analysed, and high-level customer profiles, of potential interest to on-line marketers, are derived and described in marketing terms.

Improving returns on stock investment through neural network selection

Abstract: The Artificial Neural Network (ANN) is a technique that is heavily researched and used in applications for engineering and scientific fields for various purposes ranging from control systems to artificial intelligence Its generalization powers have not only received admiration from the engineering and scientific fields, but in recent years, the finance researchers and practitioners are taking an interest in the application of ANN Bankruptcy prediction, debt-risk assessment and security market applications are the three areas that are heavily researched in the finance arena The results, this far, have been encouraging as ANN displays better generalization power as compared to conventional statistical tools or benchmark With such intensive research and proven ability of the ANN in the area of security market application and the growing importance of the role of equity securities in Singapore, it has motivated the conceptual development of this project in using the ANN in stock selection With its proven generalization ability, the ANN is able to infer from historical patterns the characteristics of performing stocks The performance of stocks is reflective of their profitability and the quality of management of the underlying company Such information is reflected in financial and technical variables As such, the ANN is used as a tool to uncover the intricate relationships between the performance of stocks and the related financial and technical variables Historical data such as financial variables (inputs) and performance of the stock (output) are used in this ANN application Experimental results obtained this far have been very encouraging

Expert system for predicting stock market timing using a candlestick chart

Abstract: It has been one of the greatest challenges to predict the stock market. Since stock prices vary dramatically, it is important to determine when to buy and sell stocks in order to get high returns from stock investment. In this study, we have developed a candlestick chart analysis expert system, or a chart interpreter, for predicting the best stock market timing. The expert system has patterns and rules which can predict future stock price movements. Defined patterns are classified into five groups with respect to their meanings: falling, rising, neutral, trend-continuation and trend-reversal patterns. The experimental results revealed that the developed knowledge base could provide excellent indicators with an average hit ratio of 72% to help investors get high returns from their stock investment. Through experiments from January 1992 to June 1997, it was proven that the developed knowledge base was time- and field-independent.

Does measuring knowledge make “cents”?

Abstract: Intellectual capital measurement is an important element of knowledge management. Organizations are grappling with the issue of how best to show that knowledge management efforts are benefiting their organization. The measurement and valuation of knowledge, especially pertaining to human capital, is an area of great interest. This article discusses this issue and proposes a valuation model for human capital.

Association, statistical, mathematical and neural approaches for mining breast cancer patterns

Abstract: Using several association and classification approaches to study breast cancer patterns, this study illustrates how these approaches can be used to predict and diagnose the occurrence of breast cancer. The results of the study, based on data obtained from a large medical facility in western Pennsylvania, show that data mining can be a viable tool for breast cancer diagnosis.

Building a rule-based machine-vision system for defect inspection on apple sorting and packing lines

Abstract: A near-infrared machine-vision system was developed for automating apple defect inspection. Fast blob extraction from fruit images was performed by using an adaptive spherical transformation. A binary decision-tree-structured rule base was established using blob feature extraction and analysis. Both off-line and on-line test results demonstrated that the rule-based system was effective for apple defect detection. Compared with the neural network method, the rules-based approach had more flexibility for changing or adding parameters, features and rules to meet various sorting requirements. The technique presented in this paper is being commercialized by a leading manufacturer of fruit/vegetable packinghouse equipment.

Multi-agent modeling of dispersed manufacturing networks

Abstract: Multi-agent modeling has emerged as a promising discipline for dealing with decision-making processes in distributed information system applications. One such application is the modeling of distributed manufacturing networks which can link up various manufacturing firms to form a “virtual” manufacturing consortium on a global basis. This paper presents a multi-agent model that can be deployed in a dispersed manufacturing network involving companies with different core competence. The multi-agent model plays the important role of monitoring the information flow and task allocation among the network companies by means of intelligent agents, which are software programs designed to accomplish specific tasks like “real” human agents with specialized skills. This proposed model is equipped with various computational intelligence technologies for various purposes, including rule-based reasoning for decomposing jobs into fundamental tasks and object-oriented technology for the design and creation of the intelligent agents, together with a genetic algorithm scheme to optimize the sequence of tasks to be carried out by various agents in order to achieve minimum overhead incurred during the operations across the dispersed network. In this paper, the detailed infrastructure of a multi-agent model and how it can be deployed to enhance the performance of a dispersed manufacturing network are presented.

An intelligent decision support system for fuzzy comprehensive evaluation of urban development

Abstract: More applications that integrate knowledge-based decision support systems, artificial neural networks, and fuzzy systems are starting to appear, and interest in such integrated systems is growing rapidly. This paper presents an integrated system in which a knowledge-based decision support system is integrated with a multilayer artificial neural network for urban development. By integrating decision support systems, knowledge-based systems, artificial neural networks, and fuzzy systems, the system achieves improvements in the implementation of each, as well as increases in the scope of the application. The paper discusses the structure of the integrated system, as well as providing an example of decision support systems application.

A Bayesian belief network-based advisory system for operational availability focused diagnosis of complex nuclear power systems

Abstract: The work reported here provides a framework of diagnostic advisory system for improved operational availability in complex nuclear power plant systems. The rule-based approach typically used for conventional expert systems is abandoned in this work. This is because of the inability of rule-based approaches to properly model the inherent uncertainties and complexities of the relationships involved in the diagnosis of actual complex engineering systems. Rather, our advisory system employs Bayesian belief network (BBN) as a high-level reasoning tool for incorporating inherent uncertainty for use in probabilistic inference. We demonstrate that a rule-based knowledge representation is simply a special case of a general BBN. First, we outline a sequential algorithm to be used in formulating the BBN-based diagnostic operational advice. Then, a prototype BBN-based representation is encoded explicitly through topological symbols and links between them, oriented in a causal direction. Once new system state related evidence from an associated sensor network is entered into this advisory system, it provides an operational advice concerning how to maintain both operational availability and safety. Based upon the framework presented here, further development of our diagnostic maintenance network, integrating a comprehensive sensor network, can be expected to lead to substantial economic gains.

Customer list segmentation using the combined response model

Abstract: Response models such as RFM (Recency, Frequency, Monetary), Logistic Regression, and Neural Networks estimate a single response model in direct marketing for segmenting and targeting customers. However, if there is considerable customer heterogeneity in the database, the models can be potentially misleading. To reflect this heterogeneity, researchers have introduced ways to combine two or more methods. Suggesting the capability of the combined model using the low correlation coefficient between them, the previous research on the combined response model did not provide answers for two important questions: (1) What are the response models that have a low correlation coefficient between them when combined? (2) Does the low correlation coefficient ensure improved performance? In this paper, we propose RFM as a method that has a low correlation coefficient when combined with Logistic Regression or Neural Networks. Our case study also concludes that the low correlation coefficient does not always ensure improved performance.

Improving the accuracy of nonlinear combined forecasting using neural networks

Abstract: The aim of the work presented in this paper is to propose artificial neural networks (ANN) as a tool for nonlinear combination of forecasts. In this study, the performance of the networks is evaluated by comparing them to three individual forecasting methods and two conventional linear combining methods. The outcome of the comparison proved that the prediction by the ANN method generally performs better than those by individual forecasting methods, as well as linear combining methods. The paper suggests that the ANN method can be used as an alternative to conventional linear combining methods to achieve greater forecasting accuracy. Meanwhile, ANNs also can be integrated with many other approaches including connectionist expert systems to improve the prediction quality further.

Evaluating the efficiency of system integration projects using data envelopment analysis (DEA) and machine learning

Abstract: Data envelopment analysis (DEA), a non-parametric productivity analysis, has become an accepted approach for assessing efficiency in a wide range of fields. Despite its extensive applications, some features of DEA remain unexploited. We aim to show that DEA can be used to evaluate the efficiency of the system integration (SI) projects and suggest the methodology which overcomes the limitation of DEA through hybrid analysis utilizing DEA along with machine learning. In this methodology, we generate the rules for classifying new decision-making units (DMUs) into each tier and measure the degree of affecting the efficiencies of the DMUs. Finally, we determine the stepwise path for improving the efficiency of each inefficient DMU.

A survey of modern knowledge modeling techniques

Abstract: A major characteristic regarding developments in the broad field of artificial intelligence (AI) during the 1990s has been an increasing integration of AI with other disciplines. A number of other computer science fields and technologies have been used in developing intelligent systems, starting from traditional information systems and databases, to modern distributed systems and the Internet. This paper surveys the knowledge modeling techniques that have received most attention in recent years among developers of intelligent systems, AI practitioners and researchers. The techniques are described from two perspectives, theoretical and practical. Hence the first part of the paper presents major theoretical and architectural concepts, design approaches, and research issues. The second part deals with several practical systems, applications, and ongoing projects that use and implement the techniques described in the first part.

IntelliSPC: a hybrid intelligent tool for on-line economical statistical process control

Abstract: Statistical process control (SPC) has become one of the most commonly used tools, for maintaining an acceptable and stable level of quality characteristics in today's manufacturing. With the movement towards a computer integrated manufacturing (CIM) environment, computer based algorithms need to be developed to implement the various SPC tasks automatically. This paper presents a hybrid intelligent tool (IntelliSPC) in which a neural network based control chart pattern recognition system, an expert system based control chart alarm interpretation system and a quality cost simulation system were integrated for on-line SPC. IntelliSPC was designed to provide the quality practitioners with the status of the process (in-control or out-of-control), the plausible causes for the out-of-control situation and cost-effective actions against the out-of-control situation. This tool was intended to be implemented in a scenario where sample data are being collected on-line by automated inspection devices and monitored by control charts. An implementation example is provided to demonstrate how the proposed hybrid system could be usefully applied in a real-world automated production line. This work confirms the potential synergies of hybrid artificial intelligence (AI) techniques in a complex problem solving procedure, such as an automated SPC scheme.

An approach to case-based system for conceptual ship design assistant

Abstract: Designers heavily depend on their experience and existing ship data, when designing a ship. In preliminary design stage especially, decision making based on the designer’s expertise and heuristic knowledge are very important factors to design process because available information is limited, and cannot be fully supported by formal design procedure and design sheet. To support these conceptual design environment, the designer’s experience and heuristic knowledge are transformed into readable formats which can be operated on computer systems. The existing ship data are very useful and important in conceptual design. To use this data efficiently, it requires basically database of the existing ships and make practical application of it. In this article, intelligent system that can be a support to the conceptual design stage based on knowledge engineering was developed. Major design factors and parameters of the existing ship data were stored case base as design cases and the case base was connected with database for information exchange among them [Brown, A., Watson, I., & Filer, N. (1995). Separating the cases from the data: towards more flexible case-based reasoning. Proc. of International Conference on Case-Based Reasoning 95 (ICCBR-95), Sesimbra in Portugal]. To extract a good and suitable design case for a new ship design from case base, learning algorithm was adapted. The obtained knowledge from designers was used to compensate for the differences between the design case and a new design. The developed interactive intelligent conceptual design system (BASCON-IV) can be applied to commercial ships and bulk carriers.

DSTM: a framework to operationalise and refine a problem solving method modeled in terms of tasks and methods

Abstract: In this paper we present DSTM, a framework that enables the operationalisation and the refinement of problem-solving methods modeled within the Task–Method paradigm. DSTM proposes an operational kernel, i.e. operational but flexible high-level constructions: modeling primitives, such as task or method, and manipulation mechanisms, such as select a method. These constructions can be customized in order to better capture the paper-based model to be operationalised. This permits the construction of an implemented system that is an explicit reification of the paper-based model, and, therefore, enables to analyze the model by means of the analysis of the system. In order to support this analysis, DSTM proposes tools that allow the knowledge-engineers to inspect the implemented model from different points of view. This facilitates the conciliation of the initial modeling phase, the model refinement phase and the operationalisation phase that are achieved when constructing a KBS.

Rule selections in fuzzy expert systems

Abstract: A fuzzy expert system consists of a database of facts and a database of rules of the type: If x is A then y is B. When a fact `x is A′' is given, a generalized modus ponens inference rule is applied to infer `y is B′'. If more than one rule is relevant, this paper suggests that a measure of nearness between A and A′ should be used and a threshold point for invoking a rule be determined in advance. This paper surveys a number of measures of nearness between fuzzy sets and recommends a `better' one for implementation.

Neural-fuzzy modeling of plastic injection molding machine for intelligent control

Abstract: Neural network and fuzzy logic reasoning can complement each other to form an integrated model which capitalizes on the merits and at the same time offsets the pitfalls of the involved computational intelligence technologies. This article presents a neural-fuzzy model which consists of a neural network for suggesting the change of process parameters, together with a fuzzy reasoning mechanism for acquiring modified parameter values based on the induced parameter values from the neural network. This model is particularly useful in parameter-based control situations where there may be multiple inputs and multiple outputs involved. This model, which serves to learn from sample data and allows to extract rules which are then fuzzified prior to fuzzy inference, is implemented for the dimensional control of injection molding parts, the dimensions of which are primarily determined by the molding process parameters such as injection time and cooling temperature.

Memory and neural network based expert system

Abstract: We suggest a hybrid expert system of memory and neural network based learning. Neural network (NN) and memory based reasoning (MBR) have common advantages over other learning strategies. NN and MBR can be directly applied to the classification and regression problem without additional transform mechanisms. They also have strength in learning the dynamic behavior of the system over a period of time. Unfortunately, they have an achilles tendon. The knowledge representation of NN is unreadable to human being and this ‘black box’ property restricts the application of NN to areas which needs proper explanations as well as precise predictions. On the other hand, MBR suffers from the feature-weighting problem. When MBR measures the distance between cases, some features should be treated more importantly than others. Although previous researchers provide several feature-weighting mechanisms to overcome the difficulty, those methods were mainly applicable only to the classification problem. In our hybrid system of NN and MBR, the feature weight set calculated from the trained neural network plays the core role in connecting both the learning strategies. Moreover, the explanation on prediction can be given by presenting the most similar cases from the case base. In this paper, we present the basic idea of the hybrid system. We also present an application example with a wafer yield prediction system for semiconductor manufacturing. Experimental results show that the hybrid system predicts the yield with relatively high accuracy and is capable of learning adaptively to changing behavior of the manufacturing system.

Knowledge architecture and system design for intelligent operation support systems

Abstract: Intelligent operation support systems emerged from the complexity of modern industrial plants and the availability of inexpensive computer hardware. Modern industrial plants often collect vast amount of process data in distributed control systems and management information systems. Time stress due to data overload and decision uncertainty increases the risk of operator errors. Comprehensive operation support to the operator in abnormal situations to reduce operator errors is strongly suggested. The solution to this problem is to design intelligent operation support systems that reduce the cognitive load placed on operators by providing guidance for knowledge-based decision making. This paper presents an intelligent operation support system (IOSS) structure using rich knowledge representation and hybrid reasoning strategy. The functional requirements and desired features for IOSS are defined. The human operator's recognition behavior is analyzed. It is shown that a hybrid reasoning environment that combines case-based reasoning (CBR), model-based reasoning (MBR) and rule-based reasoning (RBR) is consistent with operator's problem solving. A multidimensional problem solving model is proposed to incorporate these requirements and human recognition behavior. The IOSS is designed by using the problem solving model as the guide.

A computational study on design and performance issues of multi-agent intelligent systems for dynamic scheduling environments

Abstract: Multi-Agent Intelligent Systems (MAIS) are loosely-coupled network of problem solving systems that, whenever needed, work together with each other to dynamically solve problems that none of the system can individually solve Among the advantages of the MAIS, when compared to the centralized systems, are increased reliability, faster problem solving, decreased communication, and more flexibility Learning to coordinate the actions is one of the most important task in MAIS In the current research, we use a widely reported dynamic job shop scheduling simulation model that uses distributed genetic learning of job scheduling strategies (Pendharkar, PC, 1997 Doctoral Dissertation, Graduate School, Southern Illinois University at Carbondale; Pendharkar, PC, 1998 Distributed learning of objectives for adaptive scheduling (in review); Pendharkar, PC, Bhattacharyya, S, 1997 Multi-agent learning in distributed artificial intelligence Proc 2nd INFORMS Conference on Information Systems and Technology San Diego, CA, p156–163; Bhattacharyya, S, Koehler, GJ, 1997 Learning by objectives for adaptive shop-floor learning Decision Sciences (to appear) Aytug, H, Koehler, GJ, Snowdon, JL, 1994 Genetic learning of dynamic scheduling within a simulation environment, Computers and Operations Research, 21 (8), 909–925; Aytug, H, Bhattacharyya, S, Koehler, GJ, Snowdon, JL, 1994 A review of machine learning in scheduling, IEEE Transactions on Engineering Management 41 (2) ) and study the performance and design issues in multi-agents information systems for dynamic scheduling in manufacturing Among the design issue and performance issues considered in this research are coordination between agents, number of agents, and frequency of learning Our results indicate that coordination between agents, and learning frequency play a significant role in the performance of multi-agent intelligent systems

CMEOC - an expert system in the coal mining industry

Abstract: This paper introduces an expert system in the mining industry, called the Coal Mining Expert and Optimization Consultation System (CMEOC) The paper mainly describes the architecture of this system, including the knowledge base, inference engine, and relevant optimization techniques The techniques include multi-objective programming, fuzzy sets, and integer programming The application results present evidence of the usefulness of the system

A comparison of machine learning techniques with a qualitative response model for auditor’s going concern reporting

Abstract: Audit reports can take the form of a non-going concern (clean) report or Going concern (financial distress) report. If a firm is facing going concern uncertainty problems the auditor has a further choice of issuing two types of audit reports, namely the modified report or the disclaimer report. The issuance of the wrong type of report can have consequences for the auditor. Prior studies have developed models in an attempt to predict the type of audit report that should be issued to clients. However, all these studies, without exception, focused on the decision whether to issue a non-going concern report or a going concern report. The present study extends this area of research by comparing three predictive models that can help facilitate the decision on the type of going concern report that should be issued. Two of the predictive models are on based machine learning techniques (Artificial Neural Networks and Expert Systems) while the third is a qualitative model (Multiple Discriminant Analysis). The validity of the models are tested by comparing their predictive ability of the type of audit report which should be issued to the client. The results of the study indicate that the artificial neural network model has a superior predictive ability in determining the type of going concern audit report that should be issued to the client.

A model-based expert control strategy using neural networks for the coal blending process in an iron and steel plant

Abstract: Two important aspects of the control of the coal blending process in the iron and steel industry are computation of the target percentage of each type of coal to be blended and the blending of the different types in the target percentages. This paper proposes an expert control strategy to compute and track the target percentages accurately. First, neural networks, mathematical models and rule models are constructed based on statistical data and empirical knowledge on the process. Then a methodology is proposed for computing the target percentages that combines the neural networks, mathematical models and rule models and uses forward chaining and model-based reasoning. Finally, the tracking control of the target percentages is carried out by a distributed PI control scheme. The expert control strategy proposed is implemented in an expert control system that contains an expert controller and a distributed controller. The results of actual runs show that the proposed expert control strategy is an effective way to control the coal blending process.

A model-based expert control system for the leaching process in zinc hydrometallurgy

Abstract: One important step in zinc hydrometallurgy is the leaching process, which involves the dissolving of zinc-bearing material in dilute sulfuric acid to form a zinc sulfate solution. The key point in the control of the process is to determine the optimal pHs of the overflows of the continuous leach process and track them. This paper describes a model-based expert control system for the leaching process, which is being used in nonferrous metals smeltery. Specifically, steady-state mathematical models and rule models are first constructed based on the chemical reactions involved, the empirical knowledge of engineers and operators, and empirical data of the process. Then, a methodology is proposed for determining and tracking the optimal pHs with an expert control strategy based on a combination of mathematical models and rule models of the process. The results of actual runs show that the proposed control strategy is an effective way to control the leaching process.

A computational model of print-quality perception

Abstract: A system has been developed that can simulate human print quality assessments for simple prints. It consists of an image analysis system and a neural network trained in differentiating between different quality prints. Humans were used to assess the print quality of a series of images of different tones, produced by a variety of printing processes. An image analysis system was employed to collect and pre-process raw image data from the prints. A neural network employing supervised learning was then used to produce computer models of the assessments. The image analysis system and neural network models were subsequently employed to predict the observer assessments for a further set of prints that had not undergone the supervised learning procedure. In the prediction trial, the system correctly classified 23 out of 24 prints.

An approach to legal reasoning based on a hybrid decision-support system

Abstract: This paper describes argument structures to generate plausible explanations for the conclusions reached by rule-based reasoning (RBR), and provides a means of integrating with case-based reasoning (CBR). The area of application is a legal domain, for which a hybrid RBR–CBR knowledge-based system was built. An underlying object-oriented knowledge representation scheme provides a means of modelling both the structural relationships among knowledge entities (i.e. rules and cases) and the control structures among them. Legal reports of previously-decided cases are used as a knowledge source for the CBR part of the system. An argumentation facility is presented, for each predicted rule-based outcome, based on Toulmin's argument structures to provide support via justifications. The framework of similarity for the case base side is based on a model which exploits the fuzzy proximity relations . Retrieved cases are used to help the decision-maker in formulating the final outcome of a new case (whose similarity with the retrieved cases is determined from fuzzy proximity relations). The system is also capable of providing justification of the case selection process.