Evaluating Exceptions on Time Slices

Abstract

Public transport schedules contain temporal data with many regular patterns that can be represented compactly. Exceptions come as modifications of the initial schedule and break the regular patterns increasing the size of the representation. A typical strategy to preserve the compactness of schedules is to keep exceptions separately. This, however, complicates the automated processing of schedules and imposes a more complex model on applications. In this paper we evaluate exceptions by incorporating them into the patterns that define schedules. We employ sets of time slices, termed multislices, as a representation formalism for schedules and exceptions. The difference of multislices corresponds to the evaluation of exceptions and produces an updated schedule in terms of a multislice. We propose a relational model for multislices, provide an algorithm for efficient evaluating the difference of multislices, and show analytically and experimentally that the evaluation of exceptions is a feasible strategy for realistic schedules.