CASW: Context Aware Sliding window for Frequent Itemset Mining over Data Streams

In recent years, advances in both hardware and software technologies coupled with high-speed data generation has led to data streams and data stream mining. Data generation has been much faster in data stream applications and scores of data is generated in quick turnaround time. Hence it becomes obvious to perform mining, data on arrival that is usually termed as data stream mining. General frequent pattern mining methods are envisaging limitations and do not support in responding to a massive quantum of data being streamed. In order to address such limitations, data mining researchers have focused on methods for conducting more efficient and effective mining tasks by scanning a database only once. As a process of evolution, sliding window model that perform mining operations focusing on updating accumulated parts over data streams, are proposed. It is hard to consider all of the frequent patterns in data stream environment as generated patterns were remarkably increasing as data streams get extended continuously. Hence, methods for efficiently compressing patterns that are generated are essential to address the limitations. Considering the challenges and shortcoming in the earlier solutions, in this paper, focus is on incremental mining of frequent patterns from the window and a solution of CASW (Context Aware Sliding Window) is proposed. There are well defined boundaries for frequent and infrequent patterns for specific patterns. In this research article, we adapt usage 184 V. Sidda Reddy, T.V.Rao and A.Govardhn of window size change for representing conceptual drift in the information stream. An experimental study carried out on the model depicts significant developments and has affirmed that the algorithm has been designed with a more efficient system than that of existing solution.