A Data Mining Algorithm for Inducing Temporal Constraint Networks

A new approach to the problem of temporal knowledge induction from a collection of temporal events is presented. As a result, a set of frequent temporal patterns is obtained, represented following the Simple Temporal Problem (STP) formalism: a set of event types and a set of constraints describing common temporal arrangements between the events. The use of a clustering technique makes it possible to discriminate between the frequent patterns that are found in the collection.