Parallel and distributed approach for incremental closed regular pattern mining

Due to large amounts of records and dimensions are increasing in available databases, pattern mining is a challenging problem in data mining research. Good number of parallel and distributed algorithms has been proposed to mine frequent itemsets based on support threshold of itemset. But, regularity of itemset is more essential than frequency of itemset in bank transactions, network transactions and sensor data, etc. Closed itemset has gained lot of attention than frequent itemset mining in recent data mining research. Based on these considerations we propose a novel approach parallel and distributed method for incremental mining of closed regular patterns using vertical data format. Our proposed method is capable of generating local models (each node has its own database summary) as well as global model of closed regular patterns (node has the summary of whole data base). This ability permits our approach to generate high contrast closed regular itemsets, which allows examining how the data is subjective at different nodes.