A Survey on Representation for Itemsets in Association Rule Mining

Mining frequent itemset is one of the main challenges in association rule mining. The efficiency of frequent itemsets depends on the computation time and the data structure used for storing the itemsets. The data structure greatly influences the space requirement. Most of the algorithms work well for a sparse dataset. However, if the dataset is large, it becomes difficult for computation, which eventually increases the execution time. This will affect the scalability of the algorithm. With a compact and concise representation of the itemsets, the itemsets can fit in the memory and hence, do not require any I/O operations. The data structures that are mostly used are array, tree, and trie. In this paper, we present a comparison of the different data structures that are used by the mining algorithms.