TOWARDS A HIGHLY EXTENSIBLE AND FLEXIBLE MIDDLEWARE FOR DATA MINING

Abstract

Information are critical assets to an organization. In order to uncover useful information, data mining tools are used to help the organization extract any hidden patterns to assist in decision making. However, software development is always in the process of evolution in order to keep up with the current dynamic environment. Thus, supporting a predefined set of data mining techniques, data sources and reporting formats is insufficient due to frequent change of requirements and expectation from business users. In this paper, we propose the architecture of a middleware for data mining which adopts plugins mechanism to help this so-called data mining middleware achieve extensibility and flexibility by providing wide spectrum of data mining techniques, data sources and reports.