A relational query primitive for constraint-based pattern mining

As a step towards the design of an Inductive Database Sys- tem, in this paper we present a primitive for constraint-based frequent pattern mining, which represents a careful trade-o between expressive- ness and eciency Such primitive is a simple mechanism which takes a relational table in input and extracts from it all frequent patterns which satisfy a given set of user-de ned constraints. Despite its simplicity, the proposed primitive is expressive enough to deal with a broad range of interesting constraint-based frequent pattern queries,using a comprehen- sive repertoire of constraints de ned over SQL aggregates. Thanks to its simplicity, the proposed primitive is amenable to be smoothly embedded in a variety of data mining query languages and be eciencly xecuted, by the state-of-the-art optimization techniques based on pushing the var- ious form of constraints by means of data reduction.