Solving and learning soft temporal constraints; ceteris paribus statements represented as soft constraints problems

Abstract

Soft temporal constraints problems (TCSPs) allow to describe in a natural way scenarios where events happen over time and preferences are associated to event distances and durations. However, sometimes such local preferences are difficult to set, and it may be easier instead to associate preferences to some complete solutions of the problem. The Constraint Satisfaction framework combined with Machine learning techniques can be useful in this respect. Soft constraints are useful in general for manipulating preferences. In particular it is possible to approximate CP nets, a graphical representation of ceteris paribus conditional preference statements, with semiring based soft constraints problems.