Data Mining Services Research Articles

Web services discovery based on the trade-off between quality and cost of service

A Web service is an autonomous unit of application logic that provides either some business functionality or information to other applications through an Internet connection. Web services discovery is the process of finding most appropriate Web services providers needed by a Web services requestor. One of the important issues in the discovery process is for Web services providers and Web services requestors to negotiate and find an integrative solution that is optimal to both sides. Web services providers use resources to provide services to requestors in return for benefits. On the other hand, Web services requestors pay for services from providers in return for benefits as well. In this scenario, both parties should have their own cost-benefit models for making such a business decision. In convention, fixed pricing strategies are used for Internet-related business models such as online bookstores. However, these may not be suitable in some business models. For example, data mining services are always one-of-a-kind services, expensive and non-repetitive that usually require a more sophisticated business model. Therefore, sophisticated Web services providers should provide a list of trade-off alternatives between the Quality of Service (QoS) they offer and the Cost of Service (CoS) they use to requestors. In this model, the QoS relates to performance-oriented capabilities and the CoS relates to services' resource requirements. To achieve an integrative solution, both parties have to evaluate the list of QoS and CoS alternatives for obtaining an appropriate combination. One of the negotiation strategies for achieving integrative solutions is called logrolling. Traditionally logrolling in two-issue (i.e., the QoS and CoS) and two-party (i.e., the Web services provider and requestor) negotiation is defined as the exchange of loss in one issue for gain in other issues that result in an increase of the overall values for both parties. It means that each party can increase the overall value by trading the less preferred issue for the more preferred, provided that a trade-off ratio is satisfactory. This paper presents a token-based approach to quantify the QoS and CoS for achieving integrative solutions.

Distributed data mining on the grid

In many industrial, scientific and commercial applications, it is often necessary to analyze large data sets, maintained over geographically distributed sites, by using the computational power of distributed and parallel systems. The grid can play a significant role in providing an effective computational support for knowledge discovery applications. We describe a software architecture for geographically distributed high-performance knowledge discovery applications called K nowledge G rid , which is designed on top of computational grid mechanisms, provided by grid environments such as Globus. The K nowledge G rid uses the basic grid services such as communication, authentication, information, and resource management to build more specific parallel and distributed knowledge discovery tools and services. The paper discusses how the K nowledge G rid can be used to implement distributed data mining services.

A data mining analysis of RTID alarms

IBM's emergency response service provides real-time intrusion detection (RTID) services through the Internet for a variety of clients. As the number of clients increases, the volume of alerts generated by the RTID sensors becomes intractable. This problem is aggravated by the fact that some sensors may generate hundreds or even thousands of innocent alerts per day. With an eye towards managing these alerts more effectively, IBM's data mining services group analyzed a database of RTID reports. The first objective was an approach for characterizing the “normal” stream of alerts from a sensor. Using such models tuned to individual sensors, we then developed a methodology for detecting anomalies. In contrast to many popular approaches, the decision to filter an alarm out or not takes into consideration the context in which it occurred and the historical behavior of the sensor it came from. Our second objective was to identify all the different profiles of our clients. Based on their history of alerts, we discovered several different types of clients, with different alert behaviors and thus different monitoring needs. We present the issues encountered, solutions, and findings, and discuss how our results may be used in large-scale RTID operations.

Mastering Data Mining: The Art and Science of Customer Relationship Management

From the Publisher: Berry and Linoff lead the reader down an enlightened path of best practices. -Dr. Jim Goodnight, President and Cofounder, SAS Institute Inc.This is a great book, and it will be in my stack of four or five essential resources for my professional work. -Ralph Kimball, Author of The Data Warehouse Lifecycle ToolkitMastering Data MiningIn this follow-up to their successful first book, Data Mining Techniques, Michael J. A. Berry and Gordon S. Linoff offer a case study-based guide to best practices in commercial data mining. Their first book acquainted you with the new generation of data mining tools and techniques and showed you how to use them to make better business decisions. Mastering Data Mining shifts the focus from understanding data mining techniques to achieving business results, placing particular emphasis on customer relationship management.In this book, you'll learn how to apply data mining techniques to solve practical business problems. After providing the fundamental principles of data mining and customer relationship management, Berry and Linoff share the lessons they have learned through a series of warts-and-all case studies drawn from their experience in a variety of industries, including e-commerce, banking, cataloging, retailing, and telecommunications.Through the cases, you will learn how to formulate the business problem, analyze the data, evaluate the results, and utilize this information for similar business problems in different industries.Berry and Linoff show you how to use data mining to:* Retain customer loyalty* Target the right prospects* Identify new markets for products and services* Recognize cross-selling opportunities on and off the Web. Thecompanion Web site features:* Updated information on data mining products and service providers* Information on data mining conferences, courses, and other sources of information* Full-color versions of the illustrations used in the book

Toward a Data Mining Service from Large and Heterogeneous Genome Databases in GenomeNet

Today, major sites of genome database services are gathering and updating over ten millions of heterogeneous database entries. Furthermore, most of them equip retrieval functions and analysis tools for processing such data. This situation is similar to some sort of data warehouses since heterogeneous and large amount of data are gathered together for trans-database search and analysis. The GenomeNet is one of such site, which contains about 20 sorts of genome databases (GenBank, EMBL, SWISS-PROT, PDB, KEGG, etc.) and provides many services including entry retrieval and sequence interpretation. However, there are few sites which provides data mining services covering all the database entries stored in them. The main difficulty here is data preparation for mining. As pointed out recent studies in computer science, it is hard work and requires deep knowledge about mining algorithm and structure of data itself. To solve this problem, we are developing a system for assisting flexible preparation of data for mining from the whole GenomeNet (Fig. 1).

A high-performance active digital library

We describe Javaflow and Paraflow, the client and server parts of a digital library, providing high-performance data-retrieval and data-mining services, with emphasis on user interface as well as computing efficiency. Paraflow is a component model for high-performance computing, implemented as a thin layer on Message Passing Interface (MPI); it controls a heterogeneous metacomputer, allowing groups of processes (services) to work collectively and communicate with each other by parallel messaging links (channels). Javaflow is a straightforward, intuitive, loosely coupled component model for Java applets, allowing them to be combined into a GUI that can run on a thin client from a standing start. Javaflow controls Paraflow through a well-exposed text interface which is carried by authenticated telnet, while the output of the computation can be web-pages, files, or high-speed graphics. As an example, we discuss the SARA remote-sensing library, which provides public retrieval of multichannel images of the Earth, as well as private, authenticated access to a variety of image-processing services running on parallel supercomputers.