Research on a Graph-Based Algorithm

Mining association rules is an important field in data mining. The article discussed a graph-based association mining algorithm, which directly generate frequent candidate itemsets through constructing directed graphs to form association rules. But this algorithm occupy a great deal of time for checking the candidate itemsets, so an improved algorithm proposed. The improved algorithm utilize the method of judging the nodes´ io-value to cut the redundant nodes of candidate itemsets. The experimental results shows that, with smaller minimum support value,the improved algorithm effectively improve the discovery efficiency of the frequent itemsets.