Journal of Intelligence and Information Systems,
Vol. 5, No. 1, June 1999
A Study on the Development of Expert System for Selecting and Modifying Orthogonal Array in Taguchi Method
Hwan-Jong Jung, Seong-Jin Cho, and Jae-Won Lee
Vol. 5, No. 1, Page: 1 ~ 12
Keywords : expert system, Taguchi Method, Robust Design
Development of OOKS : a Knowledge Base Model Using an Object-Oriented Database
Joon-Young Huh, Hyung-Min Kim, Kun-Woo Yang, and Ji-Yun Choi
Vol. 5, No. 1, Page: 13 ~ 34
Keywords : Knowledge Base, ODBMS(Object-Oriented Database Management System)
Building a knowledge base effectively has been an important research area in the expert systems field. A variety of approaches have been studied including rules, semantic networks, and frames to represent the knowledge base for expert systems. As the size and complexity of the knowledge base get larger and more complicated, the integration of knowledge based with database technology cecomes more important to process the large amount of data. However, relational database management systems show many limitations in handing the complicated human knowledge due to its simple two dimensional table structure. In this paper, we propose Object-Oriented Knowledge Store (OOKS), a knowledge base model on the basis of a frame sturcture using an object-oriented database. In the proposed model, managing rules for inferencing and facts about objects in one uniform structure, knowledge and data can be tightly coupled and the performance of reasoning can be improved. For building a knowledge base, a knowledge script file representing rules and facts is used and the script file is transferred into a frame structure in database systems. Specifically, designing a frame structure in the database model as it is, it can facilitate management and utilization of knowledge in expert systems. To test the appropriateness of the proposed knowledge base model, a prototype system has been developed using a commercial ODBMS called ObjectStore and C++ programming language.
The Design of EDI Controls using Neural Network
Sang-Jae Lee, and In-Goo Han
Vol. 5, No. 1, Page: 35 ~ 48
Keywords : EDI, neural network, EDI controls
Many organizational contexts should be considered in designing EDI controls to make control systems effective and efficient. This paper gives a description of the neural network model for suggesting the extent of effective EDI controls for a company that has specific organizational environment. Feedforward backpropagation neural network models are designed to predict the state of 12 modes of EDI controls from the sate of environment. The predictive power of the system is compared with that of multivariate regression analysis to evaluate the effectiveness of using neural network model in predicting the level of EDI controls. The results show that the neural network model outperforms regression analysis in predictive accuracy. The controls that have high estimated value in the model are likely to be critical controls and EDI auditor or management can enhance investment of IS resources to enhance these controls.
Knowledge-Based methodologies for the Credit Rating : Application and Comparison
Seok-Chin Chu, Jae-Kyeong Kim, Tae-Kyung Sung, and Joong-Han Kim
Vol. 5, No. 1, Page: 49 ~ 64
Keywords : ID3, Meta-Learning
Case-Based Conflict Resolution in Agent-Based Collaborative Design System
kyung-Ho Lee, and Kyu-Yeul Lee
Vol. 5, No. 1, Page: 65 ~ 80
Keywords : Conflict Resolution, Agent-Based System, Case-Based Reasoning, Collaborative Work, Ship Design
Train Operations Scheduling System for Train Time-Table
Young-Hoon Yu, Jong-Gyu Hwang, and Geun-Sik Jo
Vol. 5, No. 1, Page: 81 ~ 93
A Study on Training Ensembles of Neural Networks - A Case of Stock Price Prediction
Young-Chan Lee, and Soo-Hwan Kwak
Vol. 5, No. 1, Page: 95 ~ 101
Keywords : Neural Network, Generalization Performance, Overfitting, Bumping, Bumping, Balancing
In this paper, a comparison between different methods to combine predictions from neural networks will be given. These methods are bagging, bumping, and balancing. Those are based on the analysis of the ensemble generalization error into an ambiguity term and a term incorporating generalization performances of individual networks. Neural Networks and AI machine learning models are prone to overfitting. A strategy to prevent a neural network from overfitting, is to stop training in early stage of the learning process. The complete data set is spilt up into a training set and a validation set. Training is stopped when the error on the validation set starts increasing. The stability of the networks is highly dependent on the division in training and validation set, and also on the random initial weights and the chosen minimization procedure. This causes early stopped networks to be rather unstable: a small change in the data or different initial conditions can produce large changes in the prediction. Therefore, it is advisable to apply the same procedure several times starting from different initial weights. This technique is often referred to as training ensembles of neural networks. In this paper, we presented a comparison of three statistical methods to prevent overfitting of neural network.
A Hybrid System of Joint Time-Frequency Filtering Methods and Neural Network Techniques for Foreign Exchange Rate Forecasting
Taeksoo Shin, and Ingoo Han
Vol. 5, No. 1, Page: 103 ~ 123
Keywords : R/S, Embedding Dimension, Signal Decomposition, Wavelet Transform, Wavelet Transform, Recurrent Neural Networks
Input filtering as a preprocessing method is so much crucial to get good performance in time series forecasting. There are a few preprocessing methods (i.e. ARMA outputs as time domain filters, and Fourier transform or wavelet transform as time-frequency domain filters) for handling time series. Specially, the time-frequency domain filters describe the fractal structure of financial markets better than the time domain filters due to theoretically additional frequency information. Therefore, we, first of all, try to describe and analyze specially some issues on the effectiveness of different filtering methods from viewpoint of the performance of a neural network based forecasting. And then we discuss about neural network model architecture issues, for example, what type of neural network learning architecture is selected for our time series forecasting, and what input size should be applied to a model. In this study an input selection problem is limited to a size selection of the lagged input variables. To solve this problem, we simulate on analyzing and comparing a few neural networks having different model architecture and also use an embedding dimension measure as chaotic time series analysis or nonlinear dynamic analysis to reduce the dimensionality (i.e. the size of time delayed input variables) of the models. Throughout our study, experiments for integration methods of joint time-frequency analysis and neural network techniques are applied to a case study of daily Korean won / U. S dollar exchange returns and finally we suggest an integration framework for future research from our experimental results.
An Empirical Study for the Expert's Reliance on the Knowledge from the Several Data Mining Techniques
GwangYong Kim
Vol. 5, No. 1, Page: 125 ~ 143
A Planning based Slot Assignment System for Containers
Dong-Jo Kim, and Young-Tack Park
Vol. 5, No. 1, Page: 145 ~ 166
Keywords : planner, slot assignment

Advanced Search
Date Range