Architecture of data mining server: DATAFRONT/Server

The data mining server, whose architecture is described in this paper, automatically extracts from a database some interesting features in the form of IF THEN rules. The architecture is a Web-based client-server architecture: the server side of the system consists of a mining engine, a Web-page-generating engine, a mining scenario, a database, and a huge data cache; the client side is just a Web browser. The mining engine is implemented as CORBA objects and works on distributed environments. We provide the typical scenario of data mining as a set of HTML documents that analysts can modify by themselves. The pages are generated dynamically by the Web-page-generating engine that reflect the user's input and the latest data. In our benchmark test, DATAFRONT/Server induced the characteristic rules from 1 million records, each having 10 condition fields, in 3 minutes and 45 seconds.