Global Positioning System Reference
In-Depth Information
spatial measure and the presence of spatial dimensions is optional. On the
other hand, implemented SOLAP systems rely on spatial dimensions and
do not allow the inclusion of spatial measures with corresponding spatial
aggregation functions (e.g., Spatialytics 2013b; Intelli3 2013; SAS 2013;
Scotch and Parmanto 2005).
Furthermore, since Mondrian has very few functions for conventional
measure aggregations and analysis, e.g., sum, count, min, max, distinct-
count, lacking a more advanced one, e.g., median, standard deviation,
GeoMondrian inherits this problem. This situation can be improved using
PostgreSQL/PostGIS and defi ning measures in terms of the SQL expression
that is executed in PostgreSQL and the result is sent back to (Geo)Mondrian.
These solutions may be complex and require a good (not basic) knowledge
of (Geo)Mondrian and PostgreSQL.
The situation aggravates when new programmers are required to
implement (S)OLAP cubes using (Geo)Mondrian. Although there is
documentation with basic features, many problems require advanced
knowledge and it is necessary to look for them among different support
communities. Proposed solutions are not always intuitive or easy to
understand and depend on the implementers' luck to fi nd a response for
the particular question at hand.
Another challenging issue may be the changes between different
versions. Currently, GeoMondrian is based on Mondrian 3.5, however,
there are new features in the beta version of Mondrian 4.0 including
several changes in the schema defi nition and getting closer to Microsoft
SQL Server Analysis Services features. When this new version is accepted
by the users, it will put more pressure on GeoMondrian developers to
include modifi cations.
Therefore, there is a need to join the efforts of the research community
and apply in practice many concepts that are already known in academic
works and prototypes. This could help not only to extend SOLAP server
functionalities, but also focus on other still unresolved aspects, e.g.,
improving performance of SOLAP queries considering pre-aggregations
(e.g., Pedersen and Tryfona 2001) as is done transparently for OLAP servers,
including multiple representations of spatial objects forming hierarchies,
among other issues. Furthermore, the question whether the integrated
approach of GIS and OLAP or spatial extensions of conventional OLAP
servers is more convenient in terms of performance and complexity should
be addressed.
Spatial OLAP front-end layer
Different front-end solutions for SOLAP were developed using free software
mainly based on the Mondrian server as we have already mentioned in
Search WWH ::
Custom Search