Knowledge Creation through Data Mining and the KDD Process
Knowledge discovery in databases (KDD), (and more specifically data mining) approaches knowledge creation from a primarily technology driven perspective. In particular, the KDD process focuses on how data is transformed into knowledge by identifying valid, novel, potentially useful, and ultimately understandable patterns in data (Spiegler, 2003; Fayyad et al., 1996). KDD is primarily used on data sets for creating knowledge through model building, or by finding patterns and relationships in data.
From an application perspective, data mining and KDD are often used interchangeably. Figure 1 presents a generic representation of a typical knowledge discovery process. This figure not only depicts each stage within the KDD process but also highlights the evolution of knowledge from data through information in this process as well as the two major types of data mining; namely, exploratory and predictive. whereas the last two steps (i.e., data mining and interpretation/evaluation) in the KDD process are considered predictive data mining. It is important to note in figure 1 that typically in the KDD process the knowledge component itself is treated as a homogeneous block. Given the well established multifaceted nature of the knowledge construct (Boland & Tenkasi, 1995; Malhotra, 2000; Alavi & Leidner, 2001; Schultze & Leidner, 2002; Wickramasinghe et al., 2003) this would appear to be a significant limitation or over simplification of knowledge creation through data mining as a technique and the KDD process in general.
Integrated view of the Knowledge Discovery Process (Adapted from Wickramasinghe et al, 2003)