| dc.description.abstract | The objective of the high-utility itemset mining task is to figure out the item or the
combination of items which achieve the high profits from the database. The proposed algorithm
named HUIM is considered as a advantageous tool to analyze the customer behavior. Nevertheless,
the item categories are not considered. The ML-HUI Miner algorithm was suggested in order to
solve this problem. It not only uses the HUIM task but also combine with the taxonomy to get the
group of high utility itemsets in many levels. Though the ML-HUI Miner is successful in
discovering itemsets from many abstractions level, it may not optimize for some databases. As a
result, it cannot present all the wonderful of the algorithm. This paper goes through these issues
by extending the previous algorithm with adding transaction merging process version to help
decrease not only the running time but also the memory usage. The list of items per level on
transaction in the database will be merged into ones for eliminating the mining time in similar
transaction. The experiment on the real dataset and synthesis dataset show that the improvement
of algorithms is faster than the original algorithm | en_US |
