MSMiner, a data mining tool, provides an overall data mining solution for enterprise decision-making and intelligent information processing. The system uses powerful metadata management as the scheduling center, and realizes the organic integration of data warehouse and data extraction, transformation, loading (ETL), data mining, online analytical processing (OLAP) and the seamless connection of various data mining algorithms. The system also provides a variety of data conversion rules and data mining algorithms, which comprehensively support the activities of enterprises in production, sales, marketing, financial management, enterprise decision-making and other fields, and has broad application prospects. At the same time, multi strategy data mining tools can also be applied to other key areas of the national economy, such as tax inspection, business marketing decisions, financial departments and so on.

MSMiner is developed with visual C++ 6.0 and can run on win98/win2000/win XP. Its main features are:

(1) Based on data warehouse and new metadata management, data warehouse is created according to the theme, and managed and maintained through metadata.

(2) Data extraction, conversion, loading and other preprocessing are convenient, and OLAP query is supported.

(3) It provides a variety of data mining algorithms such as decision tree, support vector machine, rough set, fuzzy clustering, case-based reasoning, statistical methods, neural computing, etc., supports data mining functions such as feature extraction, classification, clustering, prediction, association rule discovery, statistical analysis, and supports high-level decision analysis functions.

(4) It realizes a visual task editing environment and a powerful task processing engine, which can quickly and effectively realize various data conversion and data mining tasks.

(5) Good scalability. Transformation rules and mining algorithms are encapsulated and modular. The system provides an open, flexible and universal interface, enabling users to add new rules and algorithms.

(6) It is easy to carry out secondary development. Combining MSMiner with specific fields can easily, quickly, flexibly and efficiently construct practical application systems.

MSMiner is mainly composed of four subsystems, and the architecture is shown in Figure 1.

            Figure 1  MSMiner architecture