Scalable Client-Server Data Mining: Technologies, Management and User Experience

Seminar Overview - 9th December, 1998

Over recent years, data warehousing has become a widely adopted approach to decision support. As data warehousing installations have matured, attention in many companies has turned to how the information contained in warehouses can be exploited more effectively for competitive advantage. Data mining has emerged as an effective technology for discovering patterns in business data. However, until recently data mining has not been effectively deployed against large databases in a multi-tier client-server IT environment.

ESPRIT During the last two years XpertRule Software, GEHE (Germany), Lloyds TSB, the PAC, and the University of Stuttgart have been partners in an EC-supported ESPRIT project called CRITIKAL. In CRITIKAL they have developed a multi-tier data mining system based on XpertRule Software's Profiler product. This system enables the performance potential of large data warehouse platforms to be effectively exploited and provides network managers with the capability to control data security and database and network loads. The CRITIKAL system has been installed and evaluated at GEHE and Lloyds TSB.
In this seminar, the CRITIKAL project partners presented the CRITIKAL approach, the benefits that it provides, and the experience of users. In addition, major IT industry players described how their product strategies are complemented by the CRITIKAL approach.

The seminar covered the following topics:
  • Data mining as a component of business intelligence solutions.
  • A multi-tier client-server architecture for managing data mining performance, resource consumption and data security.
  • The applicability and roles of different approaches to multi-tier scalable data mining (tree induction and association rule discovery).
  • A process (methodology) for data mining.
  • The experience of end users in identifying business problems, involving business users, interfacing with existing IT infrastructure and deploying data mining results.