Linguistic Association Rules Mining

Abstract

Typical association rules partitioning techniques such as binning, statistical and evolutionary techniques have contributed to a certain extend in dealing with quantitative and categorical attribute. However, fuzzy theory has been proved as the most prominent technique in dealing with soft boundary intervals. Thus fuzzy association rules mining has become an important field in quantitative association rules mining. Since it is good in dealing with linguistic representation on soft boundary intervals, fuzzy association rules mining has emerged as one of the research focus in linguistic association rules mining. The most obvious difference in linguistic association rules mining and conventional Boolean association rules mining lies in the support value of each item. The support in Boolean association rules is the number of occurrence for a particular item in the dataset while the support in fuzzy-based linguistic association rules is the probability ratio of occurrence of the respective item in the dataset. Therefore, a different set of standard needs to be drawn spesifically for fuzzy-based linguistic association rules analysis in A priori algorithm. This paper has proposed a set of standard consisting of two heuristic rules to overcome the problem mentioned. It is important especially in frequent patterns searching and rules testing process.