Planning for Geospatial Data Integration

In the recent past there have been significant research and industrial efforts to support access to large amount of geospatial data sources. Examples of industrial efforts include different data access standards proposed by the Open Geospatial Consortium (OpenGIS)1 and the availability of satellite imagery and other data on websites, such as TerraServer2. In order to better understand the research challenges involved in designing a geospatial data integration system, I co-developed an application called the Building Finder (Michalowski et al. 2004). The Building Finder application integrates satellite imagery from TerraServer, road network information from the Tiger Line files3, and building information from various assessor’s web sites. Our experiences with the Building Finder showed that geospatial data integration involves six major research aspects: (1) accessing geospatial data, (2) modeling geospatial data sources and operations, (3) query reformulation, (4) optimization, (5) execution, and (6) visualizing the results. However, all the efforts have mainly focused on providing access to various geographic datasets and visualizing the available geospatial data. In my thesis plan to focus on modelling geospatial data sources and operations, reformulating user queries into an integration plan consisting of a set of queries on available data sources and operations, and optimizing the generated plan. In this abstract, I present my research on developing a geospatial data integration framework that can be utilized to rapidly generate applications, such as the Building Finder. Figure 1 shows the architecture of my framework. In particular, I focus on the work that I have done on modelling data sources and complex operations, query reformulation, and optimization. I conclude this abstract with a discussion of the work that remains to be done.