Extracting mutually dependent multisets

Abstract

In this paper, we extend mutually dependent patterns as itemsets introduced by Ma and Hellerstein (2001) to mutually dependent multisets allowing two or more occurrences of the same items. Then, by improving the algorithm to extract all of the mutually dependent patterns based on Apriori with maintaining itemsets and their supports, we design the algorithm to extract all of the mutually dependent multisets based on AprioriTid with traversing a database just once and maintaining both multisets and their tail occurrences but without computing overall multiplicity of items in multisets. Finally, we give experimental results to apply the algorithm to both real data as antibiograms consisting of a date, a patient id, a detected bacterium, and so on and artificial data obtained by repeating items in transaction data.