A Coherent Rule Mining Method for Incremental Datasets based on Plausibility

For traditional itemset mining techniques like Apriori and FP-Growth, multiple passes of the dataset is required to mine frequent or rare itemsets. So a clarity based rule mining algorithm is proposed which uses an interesting measure called plausibility. Plausibility is the probability of the assumed facts to be true if the conclusion is true. This proposed algorithm can mine association rules by a single pass through the file. Instead of multiple passes, a knowledge link matrix will be maintained by identifying the whole itemsets. Along with discovering the frequent itemsets, the rules too will be mined based on the plausibility measure. The main advantage of this proposed algorithm is that it will be very useful for incremental datasets. For incremental datasets the data will be always incrementing. Finding rules from these datasets is always challenging. But the single pass benefits the need to update only the matrix in case of incremental datasets. Because of its one time file access, this algorithm is supposed to consume less space compared to the FP-growth algorithm. (C) 2016 The Authors. Published by Elsevier Ltd.