In developing a spatial database management system(DBMS) for a geographic information system(GIS), we have to consider two aspects - compliance with standards and high performance of the query processing. The former comes from the lack of compatibility and portability between spatial DBMSs. Currently, many spatial DBMSs are being developed based on the object-oriented database management system(OODBMS). There was no firm standard in the area of the OODBMS, but a few efforts have been made recently. The latter results from the fact that GIS has to manage both spatial data and non-spatial data. Generally, spatial data have properties of high complexity and large size. Moreover, the operational cost of spatial data is higher than that of non-spatial data. Therefore, the spatial DBMS is required to have an ability of processing queries with high performance.
In this thesis, we design and implement a query executor for a spatial object management system called OMEGA(Object Management systEm for Geo-spatial Applications) with the goal of fulfilling above requirements. For the standard, we comply with an extended the ODMG specification. The ODMG specification defines only non-spatial features of the OODBMS, so we develop a query executor by the specification of the spatial object algebra (SOA) which defines an algebra for integrated queries which handle both spatial and non-spatial data by extending the OQL of the ODMG specification. For high performance, we use the iterator technique when implement physical operators. The iterator method is well known for its efficiency and compatibility between operators. Also, we implement spatial operators by separating the filter step and the refinement step. This enables the execution of more effective query execution plans which interleave spatial operators and non-spatial operators.
