Monotone Constraints in Frequent Tree Mining

Recent studies show that using constraints that can be pushed into the mining pro- cess, substantially improves the performance of frequent pattern mining algorithms. How- ever the constraints and algorithms have not yet been explored for frequent structure min- ing. In this paper we present monotone con- straints for trees and develop an opportunis- tic pruning algorithm that mines frequent trees with monotone constraints. We illus- trate the use of these constraints within the application of web log mining. Finally, the eect of applying these constraints on syn- thetic data sets is evaluated. The opportunis- tic pruning methods leads to a considerable speedup and a reduction in the number of candidates generated compared to the basic algorithm.