• Selected
• Entire List

Abstract:
Market segmentation is an important issue in today's intensely ompetitive environment. While many methods have been proposed for market segmentation, they can be classified into two categories: descriptive and predictive. Descriptive methods are optimized for segment identifiability while predictive methods are optimized for segment responsiveness. Most existing segmentation methods cannot effectively address both identifiability and responsiveness goals of market segmentation due to their focus on only one aspect of the multiobjective problem. This paper proposes a new market segmentation method that unifies descriptive and predictive methods by simultaneously optimizing multiple objectives. The unified market segmentation method overcomes the limitations of existing segmentation methods and generates Pareto optimal solution sets. It also suggests the optimal number-of-segments and the best solution based on the characteristics of the Pareto front. As a result, the method presents a unified view of possible segmentation solutions and automatically selects the best solution(s) by balancing the tradeoffs. We demonstrate the benefits of our method by empirically evaluating it using data from a cell phone service provider.

market segmentation, Pareto optimal solutions, multiple objective optimization

Abstract:
The World Wide Web is transitioning from being a mere collection of documents that contain useful information toward providing a collection of services that perform useful tasks. The emerging Web service technology has been envisioned as the next technological wave and is expected to play an important role in this recent transformation of the Web. By providing interoperable interface standards for application-to-application communication, Web services can be combined with component based software development to promote application interaction and integration both within and across enterprises. To make Web services for service-oriented computing operational, it is important that Web service repositories not only be well-structured but also provide efficient tools for developers to find reusable Web service components that meet their needs. As the potential of Web services for service-oriented computing is being widely recognized, the demand for effective Web service discovery mechanisms is concomitantly growing. A number of techniques for Web service discovery have been proposed, but the discovery challenge has not been satisfactorily addressed. Unfortunately, most existing solutions are either too rudimentary to be useful or too domain dependent to be generalizable.

In this paper, we propose a Web service discovery approach that combines clustering techniques with string matching and leverages the semantics of the XML-based service specification in WSDL documents. We believe that this is one of the first attempts at applying data mining techniques in the Web service discovery domain. Our proposed approach has several appealing features: (1) It minimizes the requirement of prior knowledge from both service consumers and publishers; (2) It avoids exploiting domain dependent ontologies; and (3) It is able to visualize the semantic relationships among Web services. We have developed a Web service discovery tool based on the proposed approach using an unsupervised artificial neural network and empirically evaluated the proposed approach and tool using real Web service descriptions drawn from operational Web service registries. We report on some preliminary results demonstrating the efficacy of the proposed approach.

Web Service, Self-organizing map, Clustering method, Service Discovery

Abstract:
The global adoption of RFID technology presents a number of challenges to IT architecture design. This paper identifies important requirements for RFID data provisioning at an abstract architecture level. A non-invasive architecture style is proposed to satisfy these requirements. Our proposed architecture style has the advantages of low entry barriers and high flexibility. The architecture style is used as a basis for evaluating three existing architectures for RFID data provisioning. Various architecture mismatches that could hinder the pace of RFID adoption are identified and discussed.

RFID, data provision, non-invasive architecture style

Abstract:
Cardinality constraints have been a useful and integral part of conceptual database design since the original entity-relationship (ER) model proposed by Chen. Subsequently many papers discussing classification frameworks for cardinality constraints have been proposed. Completeness of such frameworks has always been in question since well-defined criteria do not exist to evaluate them. We propose a "reverse engineering" approach, i.e., one of defining conceptual modeling constraint completeness based on mappings from the relational model. We develop a correspondence for combinations of relational algebra operators to existing semantic constraint types. In doing so, we also come up with a new category of cardinality constraints not previously examined in literature. We feel our work has useful implications from a semantics perspective, and more importantly from a practical standpoint in developing automated mechanisms for implementing cardinality constraints.

data management, conceptual modeling, cardinality, business rules, constraints, completeness, relational algebra

Abstract:
RFID-based technology for identification and tracking has potential to cut costs and provide strategic information for real-time decision support. To be useful, the generated data must be linked and distributed through the enterprise and the supply chain. We cannot achieve this kind of integration without effectively modeling the semantics of the data. Deficiencies in existing modeling constructs lead us to introduce a supertype/subtype relationship. We also discuss relational implications.

conceptual modeling, typing relationship, supertype, subtype, RFID data

Abstract:
Availability of increasingly rich biological data has tremendous potential for discovering new knowledge that is mostly found via the links among data sources. The current simple ways of implementing these links do not explicitly represent their semantics, and therefore, fail to provide useful information. In order to facilitate knowledge discovery, we believe it is important to understand the semantics of biological data entities, including biological sequences and their structures, and more importantly, the types of links among these entities. With the links formally defined and labeled, operators based on these semantics can then be developed to aid the biologist in dynamically linking the data sources, which will greatly improve the opportunities for knowledge discovery.

biological data, data links, knowledge discovery