Database Reference
In-Depth Information
The technology of Linked Data (LD) promises to remedy the above problem
by enabling data to reference other data and be linked to them. In this sense,
a user query targeting different information islands can be able to discover all
needed information by exploiting the links between these islands. However, the
big problem with LD technology is that it has not reached the performance levels
and maturity of the traditional database technology. In addition, by also consid-
ering that users are increasingly providing a geospatial aspect to the description
of the information that they publish (e.g., consider that images taken by various
devices are now tagged with geospatial information), another problem is raised
concerning the ecient evaluation of respective geospatial queries by exploiting
the LD technology.
The current LD management systems are making good progress towards solv-
ing the above problems with commercial LD engines such as Virtuoso
1
reaching
a good level of performance by exploiting different storing techniques at the
physical level (e.g., no-sql-like solutions). However, such systems are not able
to handle all possible cases in geospatial information management. In addition,
they cannot reach good performance levels for all possible types of queries. They
are also not very scalable unless the user pays a great amount of money so that
particular clustered or cloud-based versions of these systems is exploited to guar-
antee a certain scalability level. Finally, usually such systems enforce the user to
handle low-level details inherent to the LD technologies used.
In the context of the InGeoCloudS project (
www.ingeoclouds.eu
)
, a LD Man-
agement System (LDMS) has been developed which provides satisfactory query
performance levels and supplies a Linked Data Management API/service which is
language independent and hides the complexities of the LD technology from the
end-user. This API also provides particular functionality which is able to manage
not only normal but also geospatial LD in terms of storing, updating, querying
and exporting them. Moreover, this API also provides the functionality of export-
ing INSPIRE-compliant data which is a feature highly important for geospatial
data providers if we consider the current and future INSPIRE
2
directives. In this
paper, apart from analysing the functionality of the API and its most important
features, we will also describe in detail the cloud-based architecture of the LDMS
which enables it to be quite scalable and able to supply appropriate quality levels
to the normal or geospatial queries posed. The latter is proven through a thor-
ough empirical evaluation which also partially justified particular features of the
architecture proposed as well as the scalability policies enforced.
The rest of the article is structured as follows. Section
2
reviews related work.
Section
3
provides background knowledge useful for the proper comprehension
of the article and its proposed system. Sections
4
and
5
analyse the proposed
system and in particular the functional and architectural extensions made to
its previous and more limited version. Section
5
presents and discusses empirical
evaluation results with respect to the query performance of the proposed system.
Finally, Sect.
6
concludes the article and draws directions for further research.
Search WWH ::
Custom Search