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

Abstract

In this pap weprH osethr# algor53fi) LCMfrM - LCM, and LCMmaxfor mining allfrfikk# t sets, frts,fi t closed item sets, and maximal frimal t sets, ret ectively,frW tr3WWfi)H#6 databases. The main theor3##6fi contrfi#k#k4 is that we constrfi) trst shapedtr3O# erO# rO#O composed of onlyfryfikH t closed item sets, which is induced by apar5 t-child rldfik3#kOfi) defined onfr#O#3 t closed item sets. By tr verk5H therefi# in adepth-firO manner LCM finds allfrfi36W t closed item sets in polynomial time per item set, without storut prutfi#OH obtained closed item sets inmemor .MorO ver we intr duce severfi algor5fi)## techniques using thespar# and densestrfi533#fi of input data.Algor6W54 for enumerk3 #k allfr#W43 t item sets and maximal frimal t item setsar obtainedfra LCM as its var#H ts. By computational experfi## ts onrfi# work and synthetic databases tocompar their per436fi)kO to the prefik3# algor#k#fi) we found thatour algor#H#fi ar fast onlarW rr worO datasets withnatur# distr -fik3##5 such as KDD-cup2000 datasets, and many other synthetic databases.