Fuzzy association rules and the extended mining algorithms

  • Authors:

  • Guoqing Chen;Qiang Wei

  • Affiliations:

  • Department of Management Science and Engineering, School of Economics and Management Tsinghua University, Beijing 100084, China;Department of Management Science and Engineering, School of Economics and Management Tsinghua University, Beijing 100084, China

  • Venue:
  • Information Sciences—Informatics and Computer Science: An International Journal
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper focuses on the notion of fuzzy association rules that are of the form X → Y, where either X or Y is a collection of fuzzy sets. The work stems from two observations. First, in generalized association rule mining, the taxonomies concerned may not be crisp but fuzzy (e.g., "Tomato" could be regarded as both "Fruit" and "Vegetable", each at a different degree). Second, managers often refer to decision rules in terms of linguistic expressions that may or may not be the nodes of the taxonomies (e.g., "VERY 'Expensive cloth' " → "Tropical fruit"). The paper deals with the fuzziness based upon fuzzy taxonomies that reflect partial belongings among itemsets, as well as upon the extended settings for the degree of support and the degree of confidence. Apriori algorithms are extended accordingly to discover association rules across the higher-level taxonomic nodes which are fuzzy sets in general. As a result, the discovered rules are fuzzy rules. Furthermore, linguistic hedges are also incorporated in mining fuzzy rules to express more meaningful knowledge. Moreover, the extended algorithms are tested with the synthetic data, revealing similar computational complexities to that of the classical algorithm. Finally, the extended algorithms are applied to a real data set with an explanation of the semantics of discovered fuzzy rules.