Geography Reference
In-Depth Information
other open-source communities are propagating open geospatial data in a big way
(OGC
2007
;UCGIS
2004
). But the availability of numerous and such large sized
databases puts a pointer towards comprehensiveness and pervasiveness of each
dataset related to any geospatial extent on earth. It has to be determined which
dataset gives most elaborate and complete information, or whether an integration
of multiple datasets is required (Batini et al.
1986
; Beeri et al.
2004
,
2005
;Chen
et al.
2003
; Butenuth et al.
2007
). One of the essential aspects of digital mapping
and online visualization of maps is the prioritized ranking of geolocations with
respect to their attributes and this facility is available as rank columns in Natural
Earth data tables which need to be merged with other datasets for creating a
complete and exhaustive mapping example.
The online mapping systems facilitate many geospatial datasets which are used,
created, edited and maintained including the use of GeoNames as the layer for
populated places. Herein lies the challenge addressed in this paper which delves
in-depth into the completeness of three datasets OSM, NE and GeoNames and
whether these can contribute singly or integrally towards better ranking of
geolocations. In this section the paper describes the definitions of the important
terms that define the current research leading to describe the objective. The next
section elaborates on the available literature and research up to now towards
merging of geospatial data and what gaps remain those need to be filled. The
methodology for filling those gaps is discussed in the subsequent section. The
test-bed is also described. The result section puts down the success of the research.
Basically this paper deals with the success rate, accuracy and efficiency of joining
diverse open-source databases and proves the concept presently using only OSM,
NE and GeoNames and adding ranking column correctly to the final database.
Simplistically put ranking of a geolocation (any place on earth depending on
scale) is the priority associated with that place to make it visible at a particular
zoom level when viewed on a dynamic web map such as OSM or Google Map.
When the user zooms on a map then which places and attributes would be visible is
decided according to their rankings (Budak et al.
2006
; Chang and Park
2006
; Safra
et al.
2010
). Another important field in the database of featured indexed contents is
geometry (Safra and Doytsher
2006a
). Geometry is the representation of latitude-
longitude of a point with the spatial reference system taken into account. It is an
important attribute in any geospatial database as it helps to index any given point
and search it uniquely. In this paper, we have developed join algorithms under the
following assumptions. First, we assume that locations of objects are recorded as
points. More complex forms of recording locations (e.g., polygons) have been
approximated by points (e.g., by computing the geometry). Second, in each dataset,
distinct objects represent distinct real-world entities. This is a realistic assumption
for many GIS applications. Finally, we consider join algorithms that use locations
of objects, and then move on to matching with names in a range around the location.
As is evident, location-based join is a non-trivial task involving extracting elements
from multiple datasets. Understanding the factors that determine the quality of
location-based join algorithms is a basis for developing join algorithms that use all
available properties.
