LCM: An Efficient Algorithm for Enumerating Frequent Closed Item Sets.

摘要: In this paper, we propose three algorithms LCMfreq, LCM, and LCMmax for mining all frequent sets, closed item maximal respectively, from transaction databases. The main theoretical contribution is that construct treeshaped transversal routes composed of only which induced by a parent-child relationship defined on sets. By traversing the route in depth-first manner, LCM finds sets polynomial time per set, without storing previously obtained memory. Moreover, introduce several algorithmic techniques using sparse dense structures input data. Algorithms enumerating are as its variants. computational experiments real world synthetic databases to compare their performance previous algorithms, found our fast large datasets with natural distributions such KDD-cup2000 datasets, many other