Mining Frequent Closed Itemsets and Generators Over Uncertain Data

The rapid development of computer technology and the widespread use of Internet products have already led us into the era of big data, and researchers are increasingly focused on developing advanced data mining techniques to extract valuable insights from large datasets. Frequent itemset mining refers to mining the frequently occurring itemsets in data, which is a reliable method to get useful information from a large amount of data and is used on the Internet, finance, medical and other fields. But as a rule, the number of frequent itemsets extracted from data is exponential, which reduces the usefulness of mining results. To achieve the compressed representation of frequent itemsets, researchers have proposed the concept of frequent closed itemsets, which are greatly reduced in number compared to frequent itemsets and retain the integrity of the information contained in frequent itemsets. In this paper, we propose an algorithm for mining closed itemsets and generators in uncertain data, which can be divided into two steps. First, all probabilistic frequent itemsets are mined, and then, based on the probabilistic frequent itemsets, the closed itemsets and generators among them are identified. And the closed itemsets and generators are associated based on the concept of equivalence classes. The algorithm is proved to be correct and efficient after evaluation on several datasets.