The Research And Application Of Association Rules Mining Algorithms Based On Directed Itemset Graph

Data mining which is also referred as knowledge discovery in databases, means a process of finding nontrivial, extraction of implicit, pervious unknown and potential useful information from data in database. Association rules mining as an important field of data mining discover interesting relationships among attributes in those data. By reading the literature domestic and abroad, we research some problem of association rules mining algorithms,the main contexts and innovations are showed as follow:We discuss the relationship between lattice theory, formal concept analysis and association rules mining and introduc a series of definition and property of association rules mining . A new frequent itemset mining algorithms based on Directed Itemset Graph(DISG) is introduced. By storing information of frequent itemset in DISG. The problem of discovering the frequent itemset from database is transformed into the search problem of DISG . A new maximal frequent itemset mining algorithms based on DISG is introduced to discover the long frequent pattern. By using depth-first strategy, the algorithms prune the searching space by computing the frequent extension set of itemset and discover all the maximal frequent itemset efficiently.A new algorithms of mining frequent closed itemset based on DISG is introduced. By using depth-first strategy the algorithms prune the searching space by judging the property of frequent closed seedset and discover all the frequent close itemset efficiently. The mining algorithms of incremental update frequent itemset, incremental update maximal frequent itemset and incremental update frequent closed itemset are designed based on DISG,. These algorithms can efficiently utilize the result mined and discover the updated frequent itemset efficiently.The algorithms proposed in this paper is tested by using the large scale dense dataset which all show good performances. We make an application experiment with the dataset of power station and achieve some valuable information.