A Cloud-Based, Geospatial Linked Data Management System - Transactions on Large-Scale-Data-and Knowledge-Centered Systems XX - page 66

semantic indexing by exploiting a particular index type which makes use of infor-
mation extractors and annotators to semantically index documents in relational
tables. Oracle exhibits advanced data manipulation and spatial analysis features
including a whole portfolio of functions for spatial data mining, geometric unions
and intersections, and linear inferencing. It also includes a GeoRaster datatype to
store and manage image and gridded raster data and meta-data. 3D geo-spatial
data are supported to enable the storage and management of lines, surfaces,
triangulated irregular networks, point clouds, and terrain models. Such data are
indexed through R-Tree indices. Oracle considers the Whole Earth geometry
model to take into account the curvature of Earths surface when performing cal-
culations on geodetic data, supports over 30 of most commonly used distance and
area units, and provides comprehensive tools for managing CRSs and respective
projections based on the European Petroleum Survey Group model and data set.
2.2 Centralized Approaches
Some centralized approaches, such as Strabon [ 11 ], offer good geo-spatial sup-
port in terms of: (a) geo-spatial operators that can be used in GeoSPARQL
( http://www.opengeospatial.org/standards/geosparql ) queries and (b) special-
ized indices as well as query evaluation/optimization methods. Such approaches
also provide a SPARQL end-point to be used for basic RDF data management.
Moreover, they have shown good performance in evaluating particular types
of geo-spatial queries. We should highlight here that no thorough performance
evaluation of geo-spatial queries has been conducted in the cloud. This could be
utilized to assess whether cloud capabilities are indeed exploited by RDF Stores
which adopt a cloud-based architecture to become more scalable.
Strabon [ 11 ] is an open-source, geo-spatial RDF store prototype which exploits
Sesame and PostGIS. It is able to exploit both the WKT and GML vocabularies
for the representation of 2D geometries and offers functions from the OpenGIS
Simple Feature Access for SQL OGC standard to manipulate spatial literals
and provide support for multiple CRSs. The RDF triples are stored using the
one table per predicate scheme of Sesame and dictionary encoding. The spatial
literals are stored in a table with a particular schema accommodating for an id
column, which represents the unique encoding of a spatial literal based on the
mapping dictionary, as well as a value column with a PostGIS geometry data
type for storing the geometry determined by the spatial literal and a srid column
to store the original CRS of the geometry.
Parliament [ 1 ] is an open-source RDF triple store which exploits a particular
storage and indexing scheme based on linked lists and memory-mapped files.
A particular extension to Parliament has been developed to enable the indexing
of geospatial data and the evaluation of GeoSPARQL queries. Three indices are
employed for geo-spatial data: (a) a R-Tree index, (b) a temporal index, and
(c) a basic numeric index for optimizing range queries on numeric data types.
Both WKT and GML are supported for representing geometries.