Logical Aspects of Dealing with Domain Knowledge in Data Mining with Association Rules

Issue title: Rough Sets, Rule-Based Approaches, and Knowledge Representation

Guest editors: Davide Ciucci, Dominik Ślęzak and Marcin Wolski

Article type: Research Article

Authors: Rauch, Jan^*

Affiliations: University of Economics, Prague, nám. W. Churchilla 4, 130 67 Prague 3, Czech Republic. [email protected]

Correspondence: [*] Address for correspondence: University of Economics, Prague, nám. W. Churcillla 4, 130 67 Prague 3, Czech Republic. The work described here has been supported by Grants IGA 20/2013 and 29/2016 of the University of Economics, Prague.

Abstract: Association rules are introduced as general relations of two general Boolean attributes derived from columns of an analysed data matrix. Expressive power of such association rules makes possible to use various items of domain knowledge in data mining. Each particular item of domain knowledge is mapped to a set of simple association rules. Simple association rules together with their logical consequences are understood as a set of consequences of a given item of domain knowledge. Such sets of consequences are used when interpreting results of a data mining procedure. Logical deduction plays a crucial role in this approach. New results on related deduction rules are presented.

Keywords: data mining, association rules, formalization of CRISP-DM, domain knowledge, logical calculus of association rules

DOI: 10.3233/FI-2016-1420

Journal: Fundamenta Informaticae, vol. 148, no. 1-2, pp. 1-33, 2016