Towards an automated conversion between spatial data models in spatial databases

Abstract

There exist several techniques of representing spatial objects in a spatial database. Among these representation techniques, also called models, the most commonly used are the spaghetti model, the network model and the topological data model. Because each model has its pros and cons, none of them is “a one size fit all”. The choice of a model to use is based on several considerations such as the type of data to be stored, the operations that are most likely to be performed on the data and the Database Management System (DBMS) that is to be used. Therefore, there is quite often a need to convert existing data from one model to another. Doing this conversion manually would be an almost impossible and error- prone task as the operator would have to manually manipulate portions of objects the number of which could increase exponentially. This highlights the need to develop algorithms to automate this model conversion. This paper presents a first attempt to develop such algorithms. The algorithms presented here are those to convert between the two mostly supported and used models in DBMSs, namely the spaghetti and topological models. To convert from the spaghetti model to topological model, topological primitives are extracted from individual objects and references to them are used to represent the objects. The reverse operation is performed to convert from topological model to spaghetti model. In order to demonstrate the applicability of the algorithms, they have been implemented in PostgreSQL/PostGIS and a sample dataset has been converted.