kDCI: on using direct count up to the third iteration

IEEE ICDM Workshop on Frequent Itemset Mining Implementations

Claudio Lucchese ; Salvatore Orlando ; Raffaele Perego

Others  :  http://CEUR-WS.org/Vol-126/kdci.pdf
PID  :  1768

来源: CEUR

【 摘 要 】

In Apriori-like algorithms, one of the most consuming operation during the frequent itemset mining processis the candidate search. At each iteration κ the whole dataset D has to be scanned, and for each transactiont in the database, every of its subsets of length κ is generated and searched within the candidates. If a candidate is matched, it means that the transaction subsumes the candidate, and therefore its support canbe incremented by one. [first paragragh]