A Novel Approach for Mining Relevant Frequent Patterns in an Incremental Database (original) (raw)

Frequent pattern mining is emerging as a powerful tool for many business applications such as ecommerce, recommender system and the group decision support system. Many techniques have been developed to mine the frequent patterns. However, it would be the centre of attraction if the degree of importance of each item is taken into consideration.. For this reason, weighted frequent pattern mining algorithms have been suggested. But these algorithms deal only with the static databases, whereas, in reality most databases are interactive and dynamic in nature. Incremental Weighted Frequent Pattern Mining based on Frequency Descending Order (IWFP FD) tree is used to deal with the dynamic nature of the databases while pushing the weight constraints into frequent pattern mining. Branch Sorting Method (BSM) with merge sort is used to restructure the IWFP FD tree. This makes it more convenient for mining the patterns from the tree. It also makes the IWFP FD highly compact to save memory space. This tree allows mining of frequent patterns through a single pass over the database.