Showing papers in "Expert Systems With Applications in 2004"

Knowledge sharing in virtual communities: an e-business perspective

Abstract: Thanks to availability of the Internet, virtual communities are proliferating at an unprecedented rate. In-depth understanding of virtual community dynamics can help us to address critical organizational and information systems issues such as communities-of-practice, virtual collaboration, and knowledge management. In this article, we develop a virtual community activity framework, integrating community knowledge sharing activity into business activities in the form of an e-business model. We examine how the level of community knowledge sharing activity leads to virtual community outcomes and whether such community outcomes are related to loyalty toward the virtual community service provider. Based on a field survey of 77 virtual communities currently operating in Freechal.com , one of Korea's largest Internet community service providers, we found that the level of community knowledge sharing activity is related to virtual community outcomes and such outcomes are significantly associated with loyalty to the virtual community service provider. These results imply that the level of community knowledge sharing activity may be a proper proxy for the state of health of a virtual community. Implications of the findings and future virtual community research directions are discussed.

An ltv model and customer segmentation based on customer value: a case study on the wireless telecommunication industry

Abstract: Since the early 1980s, the concept of relationship management in marketing area has gained its importance. Acquiring and retaining the most profitable customers are serious concerns of a company to perform more targeted marketing campaigns. For effective customer relationship management, it is important to gather information on customer value. Many researches have been performed to calculate customer value based on Customer lifetime value (LTV). It, however, has some limitations. It is difficult to consider the defection of customers. Prediction models have focused mainly on expected future cash flow derived from customers' past profit contribution. In this paper we suggest an LTV model considering past profit contribution, potential benefit, and defection probability of a customer. We also cover a framework for analyzing customer value and segmenting customers based on their value. Customer value is classified into three categories: current value, potential value, and customer loyalty. Customers are segmented according to three types of customer value. A case study on calculating customer value and segmenting customers of a wireless communication company will be illustrated.

A resource-based perspective on knowledge management capability and competitive advantage: an empirical investigation

Abstract: The concept of knowledge management (KM) as a powerful competitive weapon has been strongly emphasized in the strategic management literature, yet the sustainability of the competitive advantage provided by KM capability is not well-explained. To fill this gap, this paper develops the concept of KM as an organizational capability and empirically examines the association between KM capabilities and competitive advantage. In order to provide a better presentation of significant relationships, through resource-based view of the firm explicitly recognizes important of KM resources and capabilities. Firm specific KM resources are classified as social KM resources, and technical KM resources. Surveys collected from 177 firms were analyzed and tested. The results confirmed the impact of social KM resource on competitive advantage. Technical KM resource is negatively related with competitive advantage, and KM capability is significantly related with competitive advantage.

Application of Web usage mining and product taxonomy to collaborative recommendations in e-commerce

Abstract: The rapid growth of e-commerce has caused product overload where customers on the Web are no longer able to effectively choose the products they are exposed to To overcome the product overload of online shoppers, a variety of recommendation methods have been developed Collaborative filtering (CF) is the most successful recommendation method, but its widespread use has exposed some well-known limitations, such as sparsity and scalability, which can lead to poor recommendations This paper proposes a recommendation methodology based on Web usage mining, and product taxonomy to enhance the recommendation quality and the system performance of current CF-based recommender systems Web usage mining populates the rating database by tracking customers' shopping behaviors on the Web, thereby leading to better quality recommendations The product taxonomy is used to improve the performance of searching for nearest neighbors through dimensionality reduction of the rating database Several experiments on real e-commerce data show that the proposed methodology provides higher quality recommendations and better performance than other CF methodologies

An integrated data mining and behavioral scoring model for analyzing bank customers

Abstract: Analyzing bank databases for customer behavior management is difficult since bank databases are multi-dimensional, comprised of monthly account records and daily transaction records. This study proposes an integrated data mining and behavioral scoring model to manage existing credit card customers in a bank. A self-organizing map neural network was used to identify groups of customers based on repayment behavior and recency, frequency, monetary behavioral scoring predicators. It also classified bank customers into three major profitable groups of customers. The resulting groups of customers were then profiled by customer's feature attributes determined using an Apriori association rule inducer. This study demonstrates that identifying customers by a behavioral scoring model is helpful characteristics of customer and facilitates marketing strategy development.

Joint optimization of customer segmentation and marketing policy to maximize long-term profitability

Abstract: With the advent of one-to-one marketing media, e.g. targeted direct mail or internet marketing, the opportunities to develop targeted marketing (customer relationship management) campaigns are enhanced in such a way that it is now both organizationally and economically feasible to profitably support a substantially larger number of marketing segments. However, the problem of what segments to distinguish, and what actions to take towards the different segments increases substantially in such an environment. A systematic analytic procedure optimizing both steps would be very welcome. In this study, we present a joint optimization approach addressing two issues: (1) the segmentation of customers into homogeneous groups of customers, (2) determining the optimal policy (i.e. what action to take from a set of available actions) towards each segment. We implement this joint optimization framework in a direct-mail setting for a charitable organization. Many previous studies in this area highlighted the importance of the following variables: R(ecency), F(requency), and M(onetary value). We use these variables to segment customers. In a second step, we determine which marketing policy is optimal using markov decision processes, following similar previous applications. The attractiveness of this stochastic dynamic programming procedure is based on the long-run maximization of expected average profit. Our contribution lies in the combination of both steps into one optimization framework to obtain an optimal allocation of marketing expenditures. Moreover, we control segment stability and policy performance by a bootstrap procedure. Our framework is illustrated by a real-life application. The results show that the proposed model outperforms a CHAID segmentation.

Mining the breast cancer pattern using artificial neural networks and multivariate adaptive regression splines

Abstract: Data mining is a very popular technique and has been widely applied in different areas these days. The artificial neural network has become a very popular alternative in prediction and classification tasks due to its associated memory characteristics and generalization capability. However, the relative importance of potential input variables and the long training process have often been criticized and hence limited its application in handling classification problems. The objective of the proposed study is to explore the performance of data classification by integrating artificial neural networks with the multivariate adaptive regression splines (MARS) approach. The rationale under the analyses is firstly to use MARS in modeling the classification problem, then the obtained significant variables are used as the input variables of the designed neural networks model. To demonstrate the inclusion of the obtained important variables from MARS would improve the classification accuracy of the networks, diagnostic tasks are performed on one fine needle aspiration cytology breast cancer data set. As the results reveal, the proposed integrated approach outperforms the results using discriminant analysis, artificial neural networks and multivariate adaptive regression splines and hence provides an efficient alternative in handling breast cancer diagnostic problems.

A purchase-based market segmentation methodology

Abstract: Market segmentation is critical for a good marketing and customer relationship management program. Traditionally, a marketer segments a market using general variables such as customer demographics and lifestyle. However, several problems have been identified and make the segmentation result unreliable. This paper develops a novel market segmentation methodology based on product specific variables such as purchased items and the associative monetary expenses from the transactional history of customers to resolve these problems. A purchase-based similarity measure, clustering algorithm, and clustering quality function are defined in this paper. A genetic algorithm approach is adopted to ensure that customers in the same cluster have the closest purchase patterns. After completing segmentation, a designated RFM model is used to analyze the relative profitability of each customer cluster. The findings from a practical marketing implementation study will also be discussed.

Investigating the role of product features in preventing customer churn, by using survival analysis and choice modeling: The case of financial services

Abstract: The enhancement of existing relationships is of pivotal importance to companies, since attracting new customers is known to be more expensive. Therefore, as part of their customer relationship management (CRM) strategy, many researchers have been analyzing ‘why’ customers decide to switch. However, despite its practical relevance, few studies have investigated how companies can react to defection prone customers by offering the right set of products. Additionally, within the current customer attention ‘hype’, one tends to overlook the nature of different products when investigating customer defection. In this research, we study the defection of the savings and investment (SI) customers of a large Belgian financial service provider. We created different SI churn behavior categories by introducing two dimensions: (i) duration of the products (fixed term versus infinity) and (ii) capital/revenue risks involved. Considering these product features, we first gain explorative insight in the timing of the churn event by means of Kaplan–Meier estimates. Secondly, we elaborate on the most alarming group of customers that emerged from the former explorative analysis. A hazard model is built to detect the most convenient product categories to cross-sell in order to reduce their churn likelihood. Complementary, a multinomial probit model is estimated to explore the customers' preferences with respect to the product features involved and to test whether these correspond with the findings of the survival analysis. The results of our study indicate that customer retention cannot be understood by solely relying on customer characteristics. In sum, it might be true that ‘not all customers are created equal’, but neither are all products.

Effective personalized recommendation based on time-framed navigation clustering and association mining

Abstract: Personalized recommendation by predicting user-browsing behavior using association-mining technology has gained much attention in web personalization research area. However, the resulting association patterns did not perform well in prediction of future browsing patterns due to the low matching rate of the resulting rules and users' browsing behavior. This research proposes a new personalized recommendation method integrating user clustering and association-mining techniques. Historical navigation sessions for each user are divided into frames of sessions based on a specific time interval. This research proposes a new clustering method, called HBM (Hierarchical Bisecting Medoids Algorithm) to cluster users based on the time-framed navigation sessions. Those navigation sessions of the same group are analyzed using the association-mining method to establish a recommendation model for similar students in the future. Finally, an application of this recommendation method to an e-learning web site is presented, including plans of recommendation policies and proposal of new efficiency measures. The effectiveness of the recommendation methods, with and without time-framed user clustering, are investigated and compared. The results showed that the recommendation model built with user clustering by time-framed navigation sessions improves the recommendation services effectively.

A scalable P2P recommender system based on distributed collaborative filtering

Abstract: Collaborative Filtering (CF) technique has been proved to be one of the most successful techniques in recommender systems in recent years. However, most existing CF based recommender systems worked in a centralized way and suffered from its shortage in scalability as their calculation complexity increased quickly both in time and space when the record in user database increases. In this article, we first propose a distributed CF algorithm called PipeCF together with two novel approaches: significance refinement and unanimous amplification, to further improve the scalability and prediction accuracy. We then show how to implement this algorithm on a Peer-to-Peer (P2P) structure through distributed hash table method, which is the most popular and efficient P2P routing algorithm, to construct a scalable distributed recommender system. The experimental data show that the distributed CF-based recommender system has much better scalability than traditional centralized ones with comparable prediction efficiency and accuracy. q 2004 Elsevier Ltd. All rights reserved.

A personalized recommender system for the cosmetic business

Abstract: In order to have an effective command of the relationship between customers and products, we have constructed a personalized recommender system which incorporates content-based, collaborative filtering, and data mining techniques. We have also introduced a new scoring approach to determine customers' interest scores on products. To demonstrate how our system works, we used it to analyze real cosmetic data and generate a recommender score table for sellers to refer to. After tracking its performance for 1 year, we have obtained quite impressive results.

Predicting the exchange traded fund DIA with a combination of genetic algorithms and neural networks

Abstract: We evaluate the performance of a heterogeneous mixture of neural network algorithms for predicting the exchange-traded fund DIA. A genetic algorithm is utilized to find the best mixture of neural networks, the topology of individual networks in the ensemble, and to determine the features set. The genetic algorithm also determines the window size of the input time-series supplied to the individual classifiers in the mixture of experts. The mixtures of neural network experts consist of recurrent back-propagation networks, and radial basis function networks. The application of genetic algorithm on the heterogeneous mixture of powerful neural network architectures shows promise for prediction of stock market time series. These highly non-linear, stochastic and highly non-stationary time series have been found to be notoriously difficult to predict using conventional linear statistical methods. In this paper, we propose a biologically inspired methodology to tackle such hard problem using a multi-faceted solution.

Integration of ART-Kohonen neural network and case-based reasoning for intelligent fault diagnosis

Abstract: This paper presents a new approach for integrating case-based reasoning (CBR) with an ART-Kohonen neural network (ART-KNN) to enhance fault diagnosis. When solving a new problem, the neural network is used to make hypotheses and to guide the CBR module in the search for a similar previous case that supports one of the hypotheses. The knowledge acquired by the network is interpreted and mapped into symbolic diagnosis descriptors, which are kept and used by the system to determine whether a final answer is credible, and to build explanations for the reasoning carried out. ART-KNN, synthesizing the theory of adaptive resonance theory and the learning strategy of Kohonen neural network, can solve the plasticity-stability dilemma of conventional neural networks. It can carry out ‘on-line’ training without forgetting previously trained patterns (stable training), and recode previously trained categories adaptive to changes in the environment and is self-organizing, which differs from most of networks that only can be carried out off-line. The proposed system has been used in the faults diagnosis of electric motor to verify the system performance. The result shows the proposed system performs better than self-organizing feature map (SOFM) based system with respect to classification rate.

Knowledge based decision making on higher level strategic concerns: system dynamics approach

Abstract: In recognizing knowledge as a new resource in gaining organizational competitiveness, knowledge management suggests a method in managing and applying knowledge for improving organizational performance. Much knowledge management research has focused on identifying, storing, and disseminating process related knowledge in an organized manner. Applying knowledge to decision making has a significant impact on organizational performance than solely processing transactions for knowledge management. In this research, we suggest a method of knowledge-based decision-making using system dynamics, with an emphasis to strategic concerns. The proposed method transforms individual mental models into explicit knowledge by translating partial and implicit knowledge into an integrated knowledge model. The scenario-based test of the organized knowledge model enables decision-makers to understand the structure of the target problem and identify its basic cause, which facilitates effective decision-making. This method facilitates the linkage between knowledge management initiatives and achieving strategic goals and objectives of an organization.

Feature-based recommendations for one-to-one marketing

Abstract: Most recommendation systems face challenges from products that change with time, such as popular or seasonal products, since traditional market basket analysis or collaborative filtering analysis are unable to recommend new products to customers due to the fact that the products are not yet purchased by customers. Although the recommendation systems can find customer groups that have similar interests as target customers, brand new products often lack ratings and comments. Similarly, products that are less often purchased, such as furniture and home appliances, have fewer records of ratings; therefore, the chances of being recommended are often lower. This research attempts to analyze customers' purchasing behaviors based on product features from transaction records and product feature databases. Customers' preferences toward particular features of products are analyzed and then rules of customer interest profiles are thus drawn in order to recommend customers products that have potential attraction with customers. The advantage of this research is its ability of recommending to customers brand new products or rarely purchased products as long as they fit customer interest profiles; a deduction which traditional market basket analysis and collaborative filtering methods are unable to do. This research uses a two-stage clustering technique to find customers that have similar interests as target customers and recommend products to fit customers' potential requirements. Customers' interest profiles can explain recommendation results and the interests on particular features of products can be referenced for product development, while a one-to-one marketing strategy can improve profitability for companies.

Customer-adapted coupon targeting using feature selection

Abstract: The management of coupon promotions is an important issue for marketing managers since it still is the major promotion medium. However, the distribution of coupons does not go without problems. Although manufacturers and retailers are investing heavily in the attempt to convince as many customers as possible, overall coupon redemption rate is low. This study improves the strategy of retailers and manufacturers concerning their target selection since both parties often end up in a battle for customers. Two separate models are built: one model makes predictions concerning redemption behavior of coupons that are distributed by the retailer while another model does the same for coupons handed out by manufacturers. By means of the feature-selection technique 'Relief-F' the dimensionality of the models is reduced, since it searches for the variables that are relevant for predicting the outcome. In this way, redundant variables are not used in the model-building process. The model is evaluated on real-life data provided by a retailer in Fast Moving Consumer Goods (FMCG). The contributions of this study for retailers as well as manufacturers are three-fold. First, the possibility to classify customers concerning their coupon usage is shown. In addition, it is demonstrated that retailers and manufacturers can stay clear of each other in their marketing campaigns. Finally, the feature-selection technique Relief-F proves to facilitate and optimize the performance of the models.

Using fuzzy cognitive map for the relationship management in airline service

Abstract: This paper proposes the usage of fuzzy cognitive map (FCM) for the management of relationships among organizational members in airline service. The main task of relationship management demands consideration of the complex causal relationship among conflict, communication, balanced power, shared values, trust, and cooperation. It is difficult even for experts in organizational behavior to cognitively predict the causal effect of one factor on the others. FCM is used to describe the inference process for the relationship management in airline service. Initially, structural equation models are used for identifying relevant relationships among the factors and indicating their direction and strength. A standardized causal coefficient is then used to create a cognitive map illustrating the effect of the status of one component on the status of another component. The cognitive map provides preliminary insights into the direction of relationship management toward maximizing effectiveness of airline service.

Segmentation of stock trading customers according to potential value

Abstract: In this article, we use three clustering methods (K-means, self-organizing map, and fuzzy K-means) to find properly graded stock market brokerage commission rates based on the 3-month long total trades of two different transaction modes (representative assisted and online trading system). Stock traders for both modes are classified in terms of the amount of the total trade as well as the amount of trade of each transaction mode, respectively. Results of our empirical analysis indicate that fuzzy K-means cluster analysis is the most robust approach for segmentation of customers of both transaction modes. We then propose a decision tree based rule to classify three groups of customers and suggest different brokerage commission rates of 0.4, 0.45, and 0.5% for representative assisted mode and 0.06, 0.1, and 0.18% for online trading system, respectively.

Application of adaptive neuro-fuzzy inference system for detection of electrocardiographic changes in patients with partial epilepsy using feature extraction

Abstract: In this study, a new approach based on adaptive neuro-fuzzy inference system (ANFIS) was presented for detection of electrocardiographic changes in patients with partial epilepsy Decision making was performed in two stages: feature extraction using the wavelet transform (WT) and the ANFIS trained with the backpropagation gradient descent method in combination with the least squares method Two types of electrocardiogram (ECG) beats (normal and partial epilepsy) were obtained from the MIT-BIH database The proposed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach Some conclusions concerning the impacts of features on the detection of electrocardiographic changes were obtained through analysis of the ANFIS The performance of the ANFIS classifier was evaluated in terms of training performance and classification accuracies and the results confirmed that the proposed ANFIS classifier has potential in detecting the electrocardiographic changes in patients with partial epilepsy

A fuzzy cognitive map approach to support urban design

Abstract: This paper is concerned with proposing a fuzzy cognitive map (FCM) driven approach for implementing expert decision support in the area of urban design. Previous research activities modeled a knowledge-based system using first order predicate calculus. The current approach utilizes the inherent analogies between SLD resolution trees and FCMs and proposes the generation of multi-branch map hierarchies in order to model the disjointed conjuncts of the logic-based rules. This research proposes specific algorithms for interpreting the logic-based rules to FCMs as well as specific algorithms and formulas for calculating the values of multi-branch map hierarchies. Finally, this paper presents preliminary experiments and comments on the usefulness of the proposed methodology tool.

Data mining to aid policy making in air pollution management

Abstract: In the past two decades, the heavy environmental loading has led to the deterioration of air quality in Taiwan. The task of controlling and improving air quality has attracted a great deal of national attention. The Taiwanese government has since set up the National Air Quality Monitoring Network (TAQMN) to monitor nationwide air quality and adopted an array of measures to combat this problem. This study applies data mining to uncover the hidden knowledge of air pollution distribution in the voluminous data retrieved from monitoring stations in TAQMN. The mining process consists of data acquisition from Web sites of 71 data gathering stations nationwide, data pre-processing using multi-scale wavelet transforms, data pattern identification using cluster analysis, and final analysis in mapping the identified clusters to geographical locations. The application of multi-scale wavelet transforms contributes greatly in removing noises and identifying the trend of data. In addition, the proposed two-level self-organization map neural network demonstrates its ability in identifying clusters on the high-dimensional wavelet-transformed space. The identified distribution of suspended particulate PM10 represents a complete, national picture of the present air quality situation, which contrasts the present pollution districts, and could serve as an important reference for government agencies in evaluating present and devising future air pollution policies.

A self-learning expert system for diagnosis in traditional Chinese medicine

Abstract: A novel self-learning expert system for diagnosis in Traditional Chinese medicine (TCM) was constructed by incorporating several data mining techniques, mainly including an improved hybrid Bayesian network learning algorithm, [email protected]?ve-Bayes classifiers with a novel score-based strategy for feature selection and a method for mining constrained association rules. The data-driven nature distinguished the system from those existing TCM expert systems based on if-then rules to address knowledge elicitation problem. Moreover, the learned knowledge was provided in multiple forms including causal diagram, association rule and reasoning rules derived from classifiers. Finally, five representative cases were diagnosed to evaluate the performance of the system and the encouraging results were obtained. The results show that the prototype system performs well in diagnosis of TCM, and could be expected to be useful in the practice of TCM.

Case-based reasoning system with Petri nets for induction motor fault diagnosis

Abstract: This paper presents an innovative approach for integrating case-based reasoning (CBR) with Petri net for the fault diagnosis of induction motors. In the CBR system, maintenance engineers can retrieve the information from previous cases which closely resemble the new problem and solve the new problem using the information from the previous cases. The proposed system has been used in fault diagnosis of electric motor to confirm the system performance. The result shows the proposed system performs better than the conventional CBR system.

An expert system for job matching of the unemployed

Abstract: This paper presents an expert system for evaluation of the unemployed at certain offered posts. The expert application uses Neuro-Fuzzy techniques for analyzing a corporate database of unemployed and enterprises profile data. The process of matching an unemployed with an offered job is performed through a Sugeno type Neuro-Fuzzy inference system. Large training sets of old historical records of unemployed (belonging to the same social class), rejected or approved at several posts, (provided by the Greek General Secretariat of Social Training) were used to define the weights of the system parameters. New instances of rejected or approved cases arriving become part of the training set. Retraining is performed after a standard amount of new cases available. The system output is a measure of the unemployed suitability for the certain job (evaluation mark).

On-line personalized sales promotion in electronic commerce

Abstract: Electronic Commerce encompasses all electronically conducted business activities, operations, and transaction processing. With the development of electronic commerce in the Internet, companies have changed the way they connect to and deal with their customers and partners. Businesses now could overcome the space and time barriers and are capable of serving customers electronically and intelligently. However, it is quite a great challenge to attract and retain the customers over Internet due to the low barrier of entrance and severe competition. Personalization, a special form of differentiation, when applied in market fragmentation can transform a standard product or service into a specialized solution for an individual. In this research, an on-line personalized sales promotion decision support system is proposed. The proposed system consists of three modules: (1) marketing strategies, (2) promotion patterns model, and (3) personalized promotion products. The marketing strategies contain sales promotion strategies and pricing strategies. Promotion patterns are generated according to various sales promotion strategies, and the promoted prices for the promotion products are generated by considering both the current stages of business life cycle and product life cycle. In the promotion patterns model, by segmenting the market, customer behaviors of three categories can be analyzed by utilizing data mining techniques and statistical analysis to generate personalized candidate promotion products. Finally, multiple evaluation indicators are used and adjusted to rank and obtain the final personalized promotion products. With the promotion products based on customers' past frequent purchase patterns, it has the potential to increase the success rate of promotion, customer satisfaction, and loyalty. In this paper, a prototype system was developed to illustrate how the proposed on-line personalized promotion decision support system works in electronic commerce and a simplified case of performance analysis was conducted for evaluation.

Better Foreground Segmentation Through Graph Cuts

Abstract: For many tracking and surveillance applications, background subtraction provides an effective means of segmenting objects moving in front of a static background. Researchers have traditionally used combinations of morphological operations to remove the noise inherent in the background-subtracted result. Such techniques can effectively isolate foreground objects, but tend to lose fidelity around the borders of the segmentation, especially for noisy input. This paper explores the use of a minimum graph cut algorithm to segment the foreground, resulting in qualitatively and quantitiatively cleaner segmentations. Experiments on both artificial and real data show that the graph-based method reduces the error around segmented foreground objects. A MATLAB code implementation is available at this http URL

Expert system of a crude oil distillation unit for process optimization using neural networks

Abstract: An expert system of crude oil distillation unit (CDU) was developed to carry out the process optimization on maximizing oil production rate under the required oil product qualities. The expert system was established using the expertise of a practical CDU operating system provided by a group of experienced engineers. The input operating variables of the CDU system were properties of crude oil and manipulated variables; while the system output variables were defined as oil product qualities. The knowledge database of the CDU operating model can be built using the input–output data with an approach of artificial neural networks (ANN). The built ANN model can be applied on predicting the oil product qualities with respect to the system input variables. In addition, a design of experiment was implemented to analyze the effect of the system input variables on the oil product qualities. Optimal operating conditions were then found using the knowledge database with an optimization method according to a defined objective function. The built expert system can provide on-line optimal operating information of the CDU process to the operators corresponding to the change of crude oil properties.

Usefulness of artificial neural networks for early warning system of economic crisis

Abstract: During the 1990s, the economic crises in many parts of the world have sparked a need in building early warning system (EWS) which produces signal for possible crisis, and accordingly various EWSs have been established In this paper, we focus on an interesting issue: 'How to train EWS?' To study this, various aspects of the training data (ie the past crisis related data) will be discussed and then several data mining classifiers including artificial neural networks (ANN) will be probed as a training tool for EWS To emphasize empirical side of the problem, EWS for Korean economy is to be constructed Our investigation suggests that ANN may be quite competitive in building EWS over other data mining classifiers

Mining class outliers: concepts, algorithms and applications in CRM

Abstract: Outliers, or commonly referred to as exceptional cases, exist in many real-world databases. Detection of such outliers is important for many applications and has attracted much attention from the data mining research community recently. However, most existing methods are designed for mining outliers from a single dataset without considering the class labels of data objects. In this paper, we consider the class outlier detection problem ‘given a set of observations with class labels, find those that arouse suspicions, taking into account the class labels’. By generalizing two pioneer contributions [Proc WAIM02 (2002); Proc SSTD03] in this field, we develop the notion of class outlier and propose practical solutions by extending existing outlier detection algorithms to this case. Furthermore, its potential applications in CRM (customer relationship management) are also discussed. Finally, the experiments in real datasets show that our method can find interesting outliers and is of practical use.