Development of SQL Interface for Spatial Data Processing in XML Databases

Some XML databases have capabilities to store spatial data with certain spatial data formats. However, spatial operations that can be handled are still limited compared to SQL databases. Some XML databases don't even have spatial operations to handle it. In this paper, a system was developed to handle the lack of spatial operations in XML databases using PostGIS extension provided by PostgreSQL by creating an SQL interface. The system transforms SQL query into an XML query (XQuery and XPath) hence the data (spatial or non-spatial) can be retrieved. The transformation depends on projection, selection, aggregation, and joining to minimize the retrieved data volume. The transformation also considers the difference between SQL and XML data structure, hence the XML element such as node, attribute, nested node (simple node and collection), and nested attribute can be retrieved using a single SQL query. Afterwards, the system rebuilds the SQL query using the retrieved data as a temporary table in the subquery and applies spatial operations to it. Finally, the query is executed to retrieve the data that has been processed using spatial operations. The system is also tested by querying large amounts of data to show the performance. The performance is measured by observing the decrease in the retrieved data volume from the XML database. The test resulted that the data volume being decreased, hence the SQL rebuild process execution time can be minimized. Based on the results and evaluation, the SQL interface has been successfully able to solve spatial operations and the difference between SQL and XML data structure problems.