Fuzzy taxonomy, quantitative database and mining generalized association rules

Article type: Research Article

Authors: Shitong, Wang^{a; b; *} | Chung, Korris F.L.^{a; b} | Hongbin, Shen^{a; b}

Affiliations: [a] School of Information, Southern Yangtse University, WuXi, Jiangsu, China | [b] Department of Computing, HongKong Polytechnic University, HongKong, China

Correspondence: [*] Corresponding author. E-mail: [email protected].

Abstract: Mining association rules from databases is still a hot topic in data mining community in recent years. Due to the existence of multiple levels of abstraction (i.e, taxonomic structures) among the attributes of the databases, several algorithms were proposed to mine generalized Boolean association rules upon all levels of presumed crisp taxonomic structures. However, fuzzy taxonomic structures may be more suitable in many practical applications. In [9], the authors proposed an approach to mine generalized Boolean association rules with such fuzzy taxonomic structures. The main contribution of this paper is to extend their idea to mine generalized association rules from quantitative databases with fuzzy taxonomic structures. A new fuzzy taxonomic quantitative database model is presented, and the experimental results on realistic databases are demonstrated to validate this new model.

Keywords: data mining, association rules, fuzzy taxonomic quantitative database

DOI: 10.3233/IDA-2005-9206

Journal: Intelligent Data Analysis, vol. 9, no. 2, pp. 207-217, 2005