Abstract:
Finding frequent itemsets from transactional data streams is a challenging task due to the large volumes of data in the data stream. It is not feasible to store all the elements of the data stream at once in the memory for future analysis. Approaches requiring multiple scans of the data stream elements are not suitable in a data stream environment. The number of frequent itemsets is too large. Searching for an itemset in a large set of itemsets is a time consuming process. In this paper an algorithm has been proposed to generate frequent closed itemsets from data stream. It generates frequent closed frequent itemsets without requiring multiple scans of the data stream elements in the sliding window. It maintains a set of closed itemsets in an incremental manner. The number of closed frequent itemsets is small as compared to that of all the frequent itemsets. These two properties of the algorithm makes it both memory and time efficient.
