Efficient search space pruning strategies for mining frequent itemsets

Enumeration of the frequent itemsets forms the computationally intensive task in Association rule mining. Pruning the search space is a primary objective in many of the proposed strategies. However some of the methods effectively apply only the downward closure property and often a number of computations are wasted in examining infrequent nodes. We propose two hybrid search strategies based on a lattice framework that apply both the upward and downward closure properties optimally to prune the search space. The strategies employ intelligent heuristics to optimally switch between a bottom up and top down phase to reduce the search space by almost fifty percent.