Analysis of different approaches for storing GML documents

Abstract

The fact that GML is an XML encoding allows it to be queried. In order to query a GML document we have designed a query language over GML/XML enriched with spatial operators. This query language has an underlying data model and algebra that supplies the semantics of the query language. In order to use this query language, it is necessary to find an implementation that allows us to exploit all its features, storing GML documents efficiently. The general aim of this paper is to study the behaviour of different alternatives over XML documents (alphanumeric data) applied to store and query GML documents (alphanumeric and spatial data). The alternatives selected use relational schemas to store GML documents because they use a complete set of data management services (including concurrency control, crash recovery, scalability, etc) and benefit from the highly optimised relational query processor. Three approaches have been used: LegoDB, a structure-mapping approach, and two simple model-mapping approaches, Monet over Relational database and XParent. We focus on the effectiveness of storage models in terms of query processing. A performance study is conducted using three data sets and the experimental results are given.