FreeSpan algorithm: the sequential pattern mining of frequent pattern projection. The core idea is the divide and conquer algorithm. The basic idea is: use frequent items to recursively project the sequence database to a smaller projection database set, and generate sub-sequence fragments in each projection database. This process divides the data and the frequent pattern set to be tested, and limits each test to a smaller projection database that matches it. Advantages: reduce the overhead required to generate candidate sequences. Disadvantages: many projection databases may be generated, which is very expensive.