Ensuring Quality of Geographic Data with UML and OCL

Abstract

Geographic data is the backbone of sophisticated applications such as car navigation systems and Geographic Information Systems (GIS). Complexity quickly arises in the production of geographic data when trying to ensure quality.We define quality as the integrity and well-formedness of the contents of the geographic data, usually enforced by external applications where constraints ensuring quality (referred to as quality constraints) are implicit, low-level and scattered throughout the application code. This has significant consequences with respect to manageability, adaptability and reuse of these constraints. This paper explains our use of UML class diagrams as conceptual model for geographic data, and how we exploited the Object Constraint Language (OCL) for describing the quality constraints in an explicit, declarative and high-level way. As our use of OCL is slightly different than it was originally intended, we present our adaptations and explain the main issues of evaluating the resulting OCL. We are confident that our specific application of OCL can be put to use in other domains where complex constraints need to be expressed in a knowledge-oriented domain.

This research,in collaboration with Tele Atlas Data Gent NV, is part of the project GeoObjects (IWT 990025), funded by the Flemish Institute for the Improvement of the Scientific-Technological Research in the Industry (IWT).