A Case Based Reasoning Framework To ExtractKnowledge From Data

Abstract

Data Mining (DM) is the search for relationships and global patterns that exist in databases, but are hidden among the vast amounts of data. Recent technologies like those embedded in DM tools, allows us to argue that knowledge can be automatically obtained from data sources. However, Knowledge Discovery in Databases (KDD) is difficult to perform. Today it is recognised that the effective discovery of new knowledge involves many tasks supported by a heterogeneous suite of tools and requires many decisions taken by experts, that must know well many DM techniques and also have a big background knowledge about the area under study. These requirements are not common to end-users; this is the reason why we propose in this article a heterogeneous architecture for knowledge extraction from data, whose main goal is to make the KDD process more accessible, and easy to perform by non-experienced users. The architecture combines the KDD process with multiple data processing technologies. Among them is a Case Based Reasoning (CBR) system intended to record the successful knowledge extraction experiments to share later. A description about each DM exercise is stored in the case base and contains details about all the tasks performed, from data selection to DM algorithms chosen, including data treatment tasks. CBR has, to date, not been applied as a management tool of KDD process, which makes this an innovative architecture. The aim of this paper is just to describe it, focusing on the CBR system. Data Mining II, C.A. Brebbia & N.F.F. Ebecken (Editors) © 2000 WIT Press, www.witpress.com, ISBN 1-85312-821-X