Data Mining Platform Design And Implementation

Data mining is a promising frontier in data and information systems and their applications. Data mining, also referred to as knowledge discovery from data, is the automated or convenient extraction of patterns representing knowledge implicitly stored in large database, date warehouse, the web, other information repositories, or data streams. Today's explosive growth of data has generated an urgent need for new automated data mining tools that can intelligently assist us in transforming the vast amount of data into useful information and knowledge.DM2 is a data mining platform designed and implemented by ourselves, the design goal of the platform is to support small and middle scale data mining projects. The DM2 platform not only supports today's popular RDBMS products, such as Oracle, MySQL, SQLServer..., but also is compatible with Weka, a famous data mining experimental system. At present, we have already implemented the platform's core, and a variety of data mining algorithms like ID3, Naive Bayes, FP-Growth, Closet... upon the basic platform infrastructure.This paper mainly introduces detail design thoughts of the DM2 platform, including the data type design, the way to interact with database, and several data mining algorithm implementations, data mining experiments on railroad transportation data with DM2 platform are also included in the paper.