Information Technology Reference
In-Depth Information
Two Revision Methods Based on Constraints:
Application to a Flooding Problem
Mahat Khelfallah and Bela ıd Benhamou
Universite de Provence, Laboratoire LSIS
39 Rue Joliot-Curie 13453 Marseille, France
{ mahat,benhamou } @cmi.univ-mrs.fr
Abstract. In this paper, we are interested in geographic information
revision in the framework of a flooding problem. We show how to express
and how to revise this problem by using simple linear constraints. We
present two revision strategies based on linear constraints resolution:
the partial revision and the global revision methods. We apply these
approaches on both a real-world flooding problem and random flooding
instances.
Keywords: Revision, Linear constraints, Geographic information.
1
Introduction
Many research works have been done in the field of knowledge revision (see [1,
5] for overviews). Revision is the restoration of the knowledge base consistency
by considering more reliable information. It identifies the inconsistencies, then
corrects them by keeping a maximum of the initial information unchanged.
In this paper, we are interested in geographic knowledge revision based on
linear constraints resolution in the framework of a flooding problem. We show
how the flooding problem is expressed by linear constraints and propose two
revision methods: the partial revision and the global revision methods.
The rest of this paper is organized as follows. We describe in section 2 the
flooding problem and show how it can be represented by linear constraints. In
section 3, the revision steps. We propose in section 4 two revision methods. We
experiment, in section 5, both revision methods on a real flooding application
and on random flooding instances. Section 6 concludes the work.
This paper is a condensed version of [4].
2
Description and Representation
of the Flooding Problem
DuringafloodingintheHerault valley (in the south of France), a part of this
area was studied in order to get correct estimates of the water heights (above the
This work has been supported by the REVIGIS project, IST-1999-14189.
Search WWH ::




Custom Search