Efficient Mining Algorithms for Inter-transactionssociation Rules
Date Issued
2007
Date
2007
Author(s)
Wang, Chen-Sheng
Abstract
In this dissertation, we propose two efficient algorithms, called ITP-Miner (Inter-Transaction Patterns Miner) and CITP-Miner (Closed Inter-Transaction Patterns Miner), for mining inter-transaction association rules. We devise three data structures for both algorithms: an ID-pair stores the information used to find inter-transaction patterns; an ITP-tree enumerates and discovers frequent inter-transaction patterns; and a CITP-tree enumerates and discovers closed frequent inter-transaction patterns.he ITP-Miner algorithm uses the ID-pairs and the ITP-tree to mine all frequent inter-transaction patterns. By using the ITP-tree, the ITP-Miner requires only one database scan and can localize joining, pruning, and support counting to a small number of ID-pairs. The experiment results show that the ITP-Miner algorithm outperforms the FITI (First Intra Then Inter) algorithm by one order of magnitude.he CITP-Miner algorithm uses the ID-pairs and the CITP-tree to mine all closed frequent inter-transaction patterns. By using the CITP-tree, the CITP-Miner can embed effective pruning strategies to avoid costly candidate generation and repeated support counting. The experiment results show that the CITP-Miner algorithm outperforms the FITI and ClosedPROWL algorithms by one order of magnitude.e also compare the ITP-Miner and CITP-Miner algorithms. Since the CITP-Miner uses effective pruning strategies for mining all closed frequent inter-transaction patterns, and the number of patterns mined by the CITP-Miner may be much less than that mined by the ITP-Miner, the experiment results show that in most of the cases, the CITP-Miner algorithm outperforms the ITP-Miner algorithm in terms of execution time, but it consumes more amount of main memory.
Subjects
Association rules
Closed itemset
Data mining
Inter-transaction association rules
File(s)![Thumbnail Image]()
Loading...
Name
ntu-96-D91725001-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):0a2e6d084bce3d068ffb427628032ed9