Efficient visualization and querying of geographic databases

Geographic Information Systems (GISs) are attracting more and more interest. For geographically distributed data, value-by-area cartograms provide a highly expressive visualization. Continuously changing CIS spatiotemporal data can be animated by cartogram animation. We propose several value-by-area cartogram animation methods, and a new algorithm for creating single value-by-area cartograms. Our algorithms provide highly expressive animations for GIS spatiotemporal databases. We describe an O(n) time approximate algorithm, which can transform a sequence of n time series data points into a linear constraint database. We also describe how the approximation enables more efficient cartogram animations, as well as more efficient evaluation of database queries. The approximate evaluation has high recall and precision on queries. We also propose a new query optimization algorithm which is based on the hypergraph representation of queries. The optimization algorithm uses recursive bi-partitioning of the hypergraph to derive an optimized query evaluation strategy. This algorithm creates evaluation strategies that are easily parallelizable, which will be more efficient in parallel computers.