ABSTRACT
ABSTRACT: In the existing mining algorithm based on uncertain data, we change the rules of building the UF-tree as follows: we merge the data item to the branch once the data item in transaction matched the tree node in a branch; otherwise, we create a new branch which is composed of the current item and subsequent items from the unmatched tree node. Then, the compressed UF-tree algorithm is proposed, while the UF-Eclat algorithm is proposed by transplanting the classic vertical mining algorithm-Eclat and applying it to uncertain data. It builds the probability vector of a single data item and calculates the degree of support for candidate data items to mine the frequent items. The results indicate that the compressed UF-tree algorithm is more efficient.
