Research On The Integration Techniques Of Distributed Geo-Spatial Data Services

Integrating spatial information via web-based geo-spatial information services is an active area of research in information integration field and has been gotton much attention recently. The research efforts have mainly concentrated on integration architecture, service description/discovery/registry mechanisms, integration models and implementation environment. However, few works have been conducted in the area of service-based spatial query, especially of the distributed spatial queries. Spatial data is in the core of spatial information systems, thus, it is very important to study the spatial data integration techniques in service oriented architecture.This dissertation addresses the geo-spatial data service integration and distributed spatial query issues under web environment from the points of spatial database and information retrieval.First, the dissertation introduces the mechanism and a logical framework of the service oriented spatial data integration, then presents the data source schema and integrated schema based on relational model. The translation procedure from the source schema to the integrated schema is discussed.Second, the dissertation addresses the global query rewriting issues of the service oriented spatial data integration. Query rewriting and optimization methods are presented based on equivalent transformation of relational operators. We study the data replication issues of spatial data services and presents strategies to deal with data fragmentation and service selection. The proposed strategies can effectively reduce the overall cost of spatial range queries.Third, the dissertation specifies the spatial join query processing based on service oriented architecture. Two strategies are proposed. The first one, called the SSJI strategy, can efficiently reduce the overall costs by precomputing and materializing the spatial semi join results. The second one partitions the join space recursively and chooses the join strategy for each pair of blocks dynamically, thus minimizes the transmission cost for different data distribution.