Efficient Mining of Time Interval-Based Association Rules

Abstract

Given market or log data, it is very useful to find two sets of items or events that occur frequently with a regular time interval. We call a time-dependent relationship between two itemsets a time interval-based association rule. Finding time interval-based association rules, however, has not been much investigated yet until now. In this paper, we propose an efficient method for finding time interval-based association rules. The proposed method transforms the original input data into a more efficient form and then utilizes the transformed data in the subsequent steps. As a result, the input/output (I/O) cost of reading the data from disk is significantly reduced. Our experiments demonstrate the efficiency of the proposed method compared with those of the existing methods.