期刊名称：International Journal of Software Engineering and Its Applications
印刷版ISSN：1738-9984
出版年度：2014
卷号：8
期号：1
页码：21-32
DOI：10.14257/ijseia.2014.8.1.02
出版社：SERSC
摘要：Frequent Pattern Tree (FP-Tree) is one of the famous data structure to keep frequent itemsets. However when the content of transactional database is modified, FP-Tree must be reconstructed again due to the changes in patterns and items support. Until this recent, most of the techniques in frequent pattern mining are using the original database to determine the items support and not from their recommended trees data structure. Therefore in this paper, we proposed a technique called Fast Determination of Item Support Technique (F-DIST) to capture the items support from our suggested Disorder Support Trie Itemset (DOSTrieIT) data structure. Experiments with the UCI datasets show that the processing time to determine the items support using F-DIST from DOSTrieIT is outperformed the classical FP-Tree technique. Furthermore, the processing time to construct a complete tree data structure for DOSTrieIT is lesser than the benchmarked CanTree data structure.
关键词：Association Rules; Frequent Pattern; Tree Structure; Fast Technique
