Text this: Best-first search–based approach for mining top-k closed frequent itemsets from uncertain databases