On the Use of Information Systems Research Methods in Datamining

Abstract

Information systems are powerful instruments for organizational problem solving through formal information processing (Lyytinen, 1987). Data mining (DM) and knowledge discovery are intelligent tools that help to accumulate and process data and make use of it (Fayyad, 1996). Data mining bridges many technical areas, including databases, statistics, machine learning, and human-computer interaction. The set of data mining processes used to extract and verify patterns in data is the core of the knowledge discovery process. Numerous data mining techniques have recently been developed to extract knowledge from large databases. The area of data mining is historically more related to AI (Artificial Intelligence), pattern recognition, statistical, and database communities, though we think there is no objective reason for that. And nowadays, although the field of data mining according to the ACM classification system* for the computing field is a subject of database applications (H.2.8) that in sequence related to database management (H.2) and to information systems field (H.), there exists a gap between the data mining and information systems communities. Each of the two scientific communities publishes its own journals and books, and organizes different conferences that rarely cover the same issues. This situation is not beneficial since both communities share in common many similar problems being solved and therefore are potentially helpful for each other. In this paper (in Section 2) we consider some existing frameworks for data mining, including database perspective and inductive databases approach, the reductionist statistical and probabilistic approaches, data compression approach, and constructive induction approach. We consider their advantages and limitations analyzing what these approaches account in the data mining research and what they do not. The study of research methods in information systems by Jarvinen (1999) encouraged us to analyse connections and appropriateness of them to the area of data mining. In Section 3 we are trying to view the data mining research as a continuous