Frequent sequence mining in large volume databases is important in many areas, e.g., biological, climate, financial databases. Exact frequent sequence mining algorithms usually read the whole database many times, and if the database is large enough, then frequent sequence mining is very long or requires supercomputers. A new probabilistic algorithm for mining frequent sequences is proposed. It analyzes a random sample of the initial database. The algorithm makes decisions
about the initial database according to the random sample analysis results and performs much faster than the exact mining algorithms. The probability of errors made by the probabilistic algorithm is estimated using statistical methods.
This work is licensed under a Creative Commons Attribution 4.0 International License.
Please read the Copyright Notice in Journal Policy.