2.3. Weighted frequent pattern mining
Since previous traditional frequent pattern mining considers
importance of all items as an equivalent level, they have
difficulty in applying various characteristics in the real world
into their own mining processes. Meanwhile, weighted frequent
pattern mining [33,14] can discover pattern results with more
usefulness by considering the different importance for each item.
As a representative weighted frequent pattern mining method,
MWAR [34] performs mining processes reflecting items’ own
weight values in order to exclude outdated items that rarely occur
in databases from its mining result. In other words, the algorithm
assigns higher weights to items that frequently occur in recent
transactions and lower weights to the ones that hardly occur in old
transactions. The item weight factor can be applied in dynamic data
stream environments. WFPMDS [35] is an algorithm for mining
weighted frequent patterns over dynamic data streams