Incremental mining maximal frequent patterns from univariate uncertain data

Recently, there has been an everyday increase in the number of sources generating univariate uncertain data. A few efficient algorithms have been proposed for maximal frequent patterns mining from univariate uncertain data statically. However, many real-life applications generate univariate uncertain databases incrementally. Obviously, it is very costly to mine maximal frequent patterns from these incremental databases using current algorithms because they must be re-run from scratch. In this paper, an incremental algorithm called IMU2P-Miner is proposed for incremental maximal frequent pattern mining from univariate uncertain data. Instead of current algorithms such as MU2P-Miner, in which the tree must be reconstructed when new data are inserted, our proposed algorithm does not need tree reconstruction, and only a path must be updated or added. To do this, an efficient tree structure, which uses a local array to keep the updates, is introduced. Therefore, it is expected that the IMU2P-Miner algorithm can be faster than current algorithms for maximal frequent patterns mining from incremental univariate uncertain databases. A comprehensive experimental evaluation is conducted by several databases to compare the performance of the proposed algorithm against the MU2P-Miner algorithm. The experimental results show that the IMU2P-Miner algorithm mines maximal frequent patterns faster than the MU2P-Miner for incremental databases.