Abstract
► Directed itemsets graph can store the information of frequent itemsets of transaction databases. ► Trifurcate linked list can meet the requirements of dense directed graph and sparse directed graph. ► TDM-MFI realizes scanning databases only once and decreases I/O resources usage. ► TDM-MFI improves storage efficiency of data structures and time efficiency of the mining method. ► TDM-MFI can produce the exact MFI and generate the high-quality association rules.
Mining maximal frequent itemsets is of paramount relevance in many of data mining applications. The “traditional” algorithms address this problem through scanning databases many times. The latest research has already focused on reducing the number of scanning times of databases and then decreasing the number of accessing times of I/O resources in order to improve the overall mining efficiency of maximal frequent itemsets of association rules. In this paper, we present a form of the directed itemsets graph to store the information of frequent itemsets of transaction databases, and give the trifurcate linked list storage structure of directed itemsets graph. Furthermore, we develop the mining algorithm of maximal frequent itemsets based on this structure. As a result, one realizes scanning a database only once, and improves storage efficiency of data structure and time efficiency of mining algorithm.