Web Service for Association Rule Mining Using XML Message Passing
Gongzhu Hu, Shaozhen Ding · Computer Applications in Industry and Engineering · 2008
Abstract Data mining tools are designed to provide the userwith operations that are easy to perform to accomplishdata mining tasks. Various data mining tools have beendeveloped, many of which provide good graphical userinterface or graphical representations of the mining results,in addition to the functionalities for performing the datamining tasks. These tools, however, are often not Web-based so they do not provide the user with universalaccessibility.In this paper, we present a Web-based data miningsystem for users to perform data mining task anywherethe Web is accessible. It is currently focused on miningassociation rules. The system uses the Web service tech-nology with communication via XML messages followingthe SOAP standard. This system is an application using theWeb services and XML technologies. 1 Introduction Data mining is a general term for data analysis tasksof discovering knowledge from large data sets. One of thedata mining tasks is to discover association rules that showthe associations (togetherness) among data items storedin the data source. Many data mining systems and toolshave been developed over the last two decades, most ofwhich are expensive commercial products, such as SASEnterprise Miner [13] and IBM Intelligent Miner [8], anda few open source systems such as Weka [16]. These sys-tems are comprehensive and powerful, but they are closedsystems operating in their own proprietary environment(both on the server and client), and do not operate in theWeb environment that increasing number of corporationsare seeking.As data mining becoming a major data analysis ap-proach today for business intelligence, many corporationshave made a lot of efforts to make data mining operationsaccessible across the company’s local network, particularlythrough the Web. Usually the employees in a companyshare a central database, while they play different roles increating and accessing the tables. Employees of differentroles are assigned different accessibilities to the database.For example, a table storing employees’ private infor-mation such as social security number and salary shouldnot be accessed by the analysts. But those tables maycontain several useful information that the analysts needto accomplish their analytical tasks such as building rules.Of course they can deliver the mining task to the personwho has the role to access the secured table. This scenariois not very efficient and may not work at all because theperson having access to the secure table may not be ableto do the data mining job. It is hence very desirable fora corporation to have a data ming system operable by allinvolved employees while leaving the database access tothe system itself. This gaol can be achieved by combiningthe data mining and Web technologies.Web service [3] is “a software system designed tosupport inter-operable machine to machine interactionover a network.” Many Web service systems use theclient/server model with communication via XMLmessages following the SOAP standard [6]. The servicesprovided by such systems often provide operations writtenin the Web Service Description Language (WSDL). TheseWeb technologies are ideal for a corporation-wide datamining system.In this paper, we present a data mining system usingWeb Services with XML massage passing as the commu-nication vehicle. The system is currently only focused onmining association rules. Users operate on the system viaa Web interface that communicates with the data miningfunctions provided by a Web server that in turn retrievesdata from an Oracle database server. The mining resultsare returned to the user as displaying and/or storing on theuser’s client machine.