Article Preview
Top1. Introduction
Data mining is the study of large quantities of data in order to find useful patterns in data. Association rule mining is a technique of data mining for finding association rules existing in the database that satisfy some minimum support and minimum confidence constraints (Agrawal & Srikant, 1994). Many researchers have focused their work on efficient mining of association rules in databases. A very influential association rule mining algorithm, Apriori, has been developed for rule mining in large transaction databases. An association rule found by apriori algorithm is of the form X→Y, where X, Y are sets of items. The support of an association rule is the ratio (in percent) of the records that contain X U Y to the total number of records in the database. For a given number of records, confidence is the ratio (in percent) of the number of records that contain X U Y to the number of records that contain X. Apriori algorithm finds all rules that meet the user-specified threshold support and confidence. Many other algorithms developed are derivative and/or extensions of this algorithm.
However, previous work has been focused on mining association rules at a single concept level. There are many applications in which mining is required at multiple levels (Kaya & Alhajj, 2004). The problem of mining association rules from transactional data was introduced by Han & Fu, 1995). For example, besides finding 70 percent of customers that purchase milk may also purchase bread, it is interesting to allow users to go to next level and show the association between toned milk and wheat bread (Rajkumar, Karthik, & Sivananda, 2003). The latter statement provides more specific and concrete information as compared to former. Therefore, a data mining system should provide efficient methods for mining multiple-level association rules (Agrawal, Imielinski, & Swami, 1993). This leads to discovery of more specific and concrete knowledge from data.