Method for mining maximal weighted frequent patterns

摘要

PURPOSE: A maximum weighting frequent pattern mining method is provided to exclude patterns having relatively less important items in the frequent patterns by weighted frequent pattern mining by avoiding repetition between frequent patterns by the maximum frequent pattern mining. CONSTITUTION: A maximum weighted frequent pattern mining device performs a first scan and second scan about transaction data(S101,S103). The device generates a weighted frequent pattern tree in a preference descending order with the second scan information(S105). The device stores conditional data required for weighted frequent pattern arrangement using the scan information(S107). The device recursively performs mining in a pattern growth method about the conditional data(S109). The device extracts the maximum weighted frequent pattern group(S111). [Reference numerals] (AA) Start; (BB) End; (S101) First scan; (S103) Second scan; (S105) Generate a weighted frequent pattern tree in a preference descending order; (S107) Store weighted frequent pattern arrangement; (S109) Recursively perform mining by using a pattern growth method; (S111) Extract the maximum weighted frequent pattern group