Fuzzy Disjunctive Temporal Problems with Classes

Abstract

This paper describes a framework for temporal reasoning that allows managing a restricted form of disjunctive temporal constraints without making the modelled problems intractable as in the case of general DTPs. This is obtained by assigning classes to the constraints and by allowing only one constraint per class, in order to build a collection of independent STPs that can share sub-problems and therefore allows increasing algorithm efficiency. The model proposed is directly applied to fuzzy constraint satisfaction problems and can be solved using an extended fuzzy path-consistency algorithm, also presented in the paper. A simple application to medical diagnosis shows its expressive power over previous tractable temporal reasoning models.