Efficient Incremental Mining Of Frequent Sequence Generators

Recently, mining sequential patterns, especially closed sequential patterns and generator patterns, has attracted much attention from both academic and industrial communities. In recent years, incremental mining of all sequential patterns (all closed sequential patterns) has been widely studied. However, to our best knowledge, there has not been any study for incremental mining of sequence generators. In this paper, by carefully examining the existing expansion strategies for mining sequential databases, we design a GenTree structure to keep track of the relevant mining information, and propose an efficient algorithm, IncGen, for incremental generator mining. We have conducted thorough experiment evaluation and the experimental results show that the IncGen algorithm outperforms state-of-the-art generator-mining method FEAT significantly.