Global Positioning System Reference
In-Depth Information
support systems (Scotch and Parmanto 2005), or (3) may be discontinued,
e.g., SOLAPLayer (Spatialytics 2013c).
The prototypes and customized solutions may confi rm the veracity of the
concepts and the feasibility to implement the proposed solutions; however,
since they are not publicly available for use, they cannot be considered as an
option for implementing the SOLAP system. Furthermore, another aspect
undertaken by several authors (e.g., Badard et al. 2012; Fidalgo et al. 2003;
2010; Dubé et al. 2009) is the need to have a SOLAP server separate from the
client application that allows cube expression in XML format. Additionally,
Dubé et al. (2009) refer to different SOLAP characteristics: (1) representation
of data and metadata, (2) usage of existing standards, (3) independence of
particular implementation in OLAP and GIS products, (4) incorporation of
calculated measures and members expressed in the schema and within a
query, and (5) separation of contents and presentation. Therefore, instead
of integrating two systems in the user's interface, another approach can
be taken by providing the management of spatial and conventional data
at the lower, SOLAP server layer, along with the help of a spatial database
layer (Baltzer 2011). Some examples are GeoMondrian (Spatialytics 2013b)
or GOLAPE (Silva et al. 2006) servers that directly interface with the
Mondrian OLAP server and extend its capabilities to handle spatial data.
These SOLAP servers, in turn, use a spatially enabled DBMS, e.g., PostGIS,
as data storage.
GeoMondrian (Spatialytics 2013b) and GOLAPE (Silva et al. 2006)
by extending Mondrian with spatial features provide XML for defi ning
multidimensional structures. However, to the best of our knowledge only
GeoMondrian is available for free. It is relatively new and still lacks some
features. For example, although it allows users to defi ne geometries for
each spatial level in the XML schema, this extension does not apply to the
measures, i.e., spatial measures and their aggregation functions cannot be
included in the XML schema. The only possibility to create spatial measures
is by specifying calculated measures “on the fl y” using the
with member
clause, as we saw in the previous examples. GeoMondrian also limits the
spatial aggregation function to spatial union missing other functions, e.g.,
intersection or difference that could be considered during roll-up operations
(Malinowski and Zimányi 2008). This situation represents a clear gap
between the research community and practitioners. On the one hand, the
research community proposes models with spatial measures (e.g., Bimonte
et al. 2010; Malinowski and Zimányi 2008; Rivest et al. 2005; Stefanovic et
al. 2000) and even a prototype with a sophisticated classifi cation of spatial
measures and aggregation operations was developed (Silva, Times et al.
2008). Other authors (e.g., Damiani and Spccapietra 2006) take yet a stronger
position indicating that multidimensional spatial models must include a
Search WWH ::
Custom Search