Discovering Critical Factors Affecting RDF Stores Success
195
Each of these solutions may be suitable and usable for some kinds of tasks and
not for others, while a one-size-fits-all killer application for this type of solutions
is not available. Under these conditions, the large number of available solutions to
handle RDF data and also the lack of valid benchmarks for their rigorous evaluation
make not trivial the task of selection of a valid RDF store during the design of a
Semantic infrastructure. In order to choose the most suitable database to be used
within a use case scenario, it is important to know the offered features, and related
advantages and drawbacks. The main objective of this work is to identify criteria that
help organizations in their evaluation, selection, and adoption of an RDF store.
To achieve this objective, it is introduced in this paper an empirical approach which
allows to gain an understanding of the main quality characteristics of RDF stores
and thus to discover, in different application domains, their critical success factors in
the role of backbone of a semantic-based architecture. In particular, according to the
defined requirements and needs of three real different self-conducted case studies,
these factors are elicited and their relative criticality is evaluated. Afterwards, each
of them is analyzed in terms of an up-to-date state of the art based on the literature
review.
The remainder of this paper is structured as follows. Section 2 reviews the literature
related to this research study, whereas Sect. 3 introduces the methodological approach
used to rank the RDF stores features. In addition, Sect. 3 illustrates the application
of the proposed methodology. Finally, Sect. 4 draws the conclusions, summarizing
the main outcomes.
2 Related Works
The evaluation of RDF stores has been recently studied in various research works. In
particular, a core topic has been their comparison in terms of performance, on the basis
of specific metrics such as query duration. In this regard, several benchmarks based on
the evaluation of these metrics (such as LUBM [4, 13], and Linked Data Benchmark
[5], etc.) have been defined and formalized. However, a framework for the definitive
and rigorous comparison of RDF store is still missing, since the current proposed
solutions only partially meet all the needed expectations for a valid benchmark (i.e.
verifiable, fair, repeatable, relevant, and economical) [5].
Aside the performances, various other attributes influence the success of an RDF
store acting as backbone of a complex semantic-based architecture. In fact, like
for other Information Systems, organizations might have different requirements and
expectations of an RDF store, depending on the specific use case scenario. This is
the reason why it is essential to pair the quantitative analysis of the RDF store performance with a qualitative analysis of its features. In this regard, Haslhofer et al.
combined in [15] a qualitative and quantitative evaluation of a set of prominent triple
stores leveraging various selected criteria. Modoni et al. [25] presented a qualitative analysis and comparison of six different triple store solutions in terms of their
capability to support streaming and security. Another interesting evaluation has been
195
Each of these solutions may be suitable and usable for some kinds of tasks and
not for others, while a one-size-fits-all killer application for this type of solutions
is not available. Under these conditions, the large number of available solutions to
handle RDF data and also the lack of valid benchmarks for their rigorous evaluation
make not trivial the task of selection of a valid RDF store during the design of a
Semantic infrastructure. In order to choose the most suitable database to be used
within a use case scenario, it is important to know the offered features, and related
advantages and drawbacks. The main objective of this work is to identify criteria that
help organizations in their evaluation, selection, and adoption of an RDF store.
To achieve this objective, it is introduced in this paper an empirical approach which
allows to gain an understanding of the main quality characteristics of RDF stores
and thus to discover, in different application domains, their critical success factors in
the role of backbone of a semantic-based architecture. In particular, according to the
defined requirements and needs of three real different self-conducted case studies,
these factors are elicited and their relative criticality is evaluated. Afterwards, each
of them is analyzed in terms of an up-to-date state of the art based on the literature
review.
The remainder of this paper is structured as follows. Section 2 reviews the literature
related to this research study, whereas Sect. 3 introduces the methodological approach
used to rank the RDF stores features. In addition, Sect. 3 illustrates the application
of the proposed methodology. Finally, Sect. 4 draws the conclusions, summarizing
the main outcomes.
2 Related Works
The evaluation of RDF stores has been recently studied in various research works. In
particular, a core topic has been their comparison in terms of performance, on the basis
of specific metrics such as query duration. In this regard, several benchmarks based on
the evaluation of these metrics (such as LUBM [4, 13], and Linked Data Benchmark
[5], etc.) have been defined and formalized. However, a framework for the definitive
and rigorous comparison of RDF store is still missing, since the current proposed
solutions only partially meet all the needed expectations for a valid benchmark (i.e.
verifiable, fair, repeatable, relevant, and economical) [5].
Aside the performances, various other attributes influence the success of an RDF
store acting as backbone of a complex semantic-based architecture. In fact, like
for other Information Systems, organizations might have different requirements and
expectations of an RDF store, depending on the specific use case scenario. This is
the reason why it is essential to pair the quantitative analysis of the RDF store performance with a qualitative analysis of its features. In this regard, Haslhofer et al.
combined in [15] a qualitative and quantitative evaluation of a set of prominent triple
stores leveraging various selected criteria. Modoni et al. [25] presented a qualitative analysis and comparison of six different triple store solutions in terms of their
capability to support streaming and security. Another interesting evaluation has been