Mining Probabilistic Generalized Frequent Itemsets in Uncertain Databases

Researchers have recently defined and presented the theoretical con-cepts and an algorithm necessary for mining so-called probabilistic frequent itemsets in uncertain databases—based on possible world semantics. Further, there exist algorithms for mining so-called generalized itemsets in certain databases, where a taxonomy ex-ists relating concrete items to abstract (generalized) items not in the database. Currently, no research has been done in formulat-ing a theory and algorithm for mining generalized itemsets from uncertain databases. Using probability theory and possible world semantics, we formulate a method for calculating the probability a generalized item will occur within an uncertain transaction