A New Compact Structure to Extract Frequent Itemsets
Discovery of association rules is an important problem in KDD process. In this paper, the authors propose a new algorithm for fast frequent itemset mining, which scan the transaction database only once. All the frequent itemsets can be efficiently extracted in a single database pass. To attempt this objective, they define a new compact data structure, called ST-Tree (Signature Transaction Tree), and a new mining algorithm ST-Mine to extract frequent itemsets. Originally introduced by Agrawal in the context of transactional databases, the association rule mining approach is now used extensively to find associations in biological databases, web log data, telecommunications data, census data and many other types of databases.
Provided by: Science and Development Network (SciDev.Net) Topic: Data Management Date Added: Dec 2011 Format: PDF