Geospatial Visual Analytics Belongs to Database Systems: The BABYLON approach (extended abstract)

Published in International Conference on Advances in Geographic Information Systems, SIGSPATIAL, 2017

Abstract:

Babylon has the following main contributions: (1) Babylon encapsulates the main steps of the geospatial map visualization process, e.g., pixelize, spatially aggregate, and render, into a set of massively parallelized GeoViz operators and the user can assemble any customized styles. (2) Babylon proposes a GeoViz-aware spatial partitioning operator that accommodates the need for visual analytics but also takes into account the load balance when processing skewed geospatial data in parallel. (3) Babylon employs a set of GeoViz query operators that extend classic spatial query operators to perform spatial queries, e.g., spatial range and join, on hybrid inputs including spatial objects and pixels. (4) Babylon provides the end users with a declarative GeoViz language to clearly describe the ultimate visualization e‚ect. Babylon optimizer co-optimizes GeoViz query and viz operators side by side and selects the best execution plan (in terms of total run time). To achieve that, the optimizer calculates the spatial distribution of the input datasets, analyzes the GeoViz workload information, and €nally proposes a GeoViz execution plan.