Data mining and data warehousing are two key technologies which have made significant contributions to the field of knowledge discovery in a variety of domains. More recently, the integrated use of traditional data mining techniques such as clustering and pattern recognition with data warehousing technique of Online
Analytical Processing (OLAP) have motivated diverse research areas for leveraging knowledge discovery from complex real-world datasets. Recently, a number of such integrated methodologies have been proposed to extract knowledge from datasets but most of these methodologies lack automated and generic methods for schema generation and knowledge extraction. Mostly data analysts need to rely on domain specific knowledge and have to cope with technological constraints in order to discover knowledge from high dimensional datasets. In this paper we present a generic methodology which incorporates semi-automated knowledge extraction methods to provide data-driven assistance towards knowledge discovery. In particular, we provide a method for constructing a binary tree of hierarchical clusters and annotate each node in the tree with significant numeric variables. Additionally, we propose automated methods to rank nominal variables and to generate candidate multidimensional schema with highly significant dimensions. We have performed three case studies on three real-world datasets taken from the UCI machine learning repository in order to validate the generality and applicability of our proposed methodology.
|Cite as: Usman, M., Pears, R. and Fong, A.C.M. (2012). Data Guided Approach to Generate Multi-dimensional Schema for Targeted Knowledge Discovery. In Proc. Data Mining and Analytics 2012 (AusDM 2012) Sydney, Australia. CRPIT, 134. Zhao, Y., Li, J. , Kennedy, P.J. and Christen, P. Eds., ACS. 229 - 240 |
(local if available)