ILUNA: Single-pass incremental method for uncertain frequent pattern mining without false positives

Abstract

Nowadays, due to the mass production of uncertain data, numerous methods have been proposed for mining frequent patterns from uncertain data; however, none of them are proper for dynamic data environments. In many real-world applications, transactions are constantly being updated. After incremental updates, the validity of the uncertain patterns changes. The existing static algorithms to handle this state have to rerun the whole mining process from scratch, which is very costly. Incremental-CUF-growth is a method dealing with dynamic data but it generates many false positives and requires an additional time-consuming database scan to filter them. To handle these drawbacks, in this paper, an efficient single-pass method called ILUNA is proposed for incremental mining of uncertain frequent patterns without false positives. It introduces two new data structures namely IUP-List and ICUP-List to efficiently store data which can be increased. Upon receiving each new database, it only updates the lists without having to rebuild them from scratch. This is the first study in which single-pass incremental mining of uncertain frequent patterns is performed. Comprehensive experimental results show that the proposed method dramatically reduces the runtime and enhances the scalability compared to the state-of-the-art methods for dense and sparse incremental datasets.