Database Reference
In-Depth Information
11.7 FUTURE DIRECTIONS
Finally, let us look into the future and make some suggestions for what may be to
come at the crossroads between GI, Linked Data, and the Semantic Web. In doing so,
though, bear in mind what was said by Niels Bohr, the great physicist: “Prediction is
very difficult, especially about the future.” To confirm the difficulty with prediction
we write this topic as new technologies are arising at a seemingly ever increasing rate.
At the time of writing the industry has seen much interest in NOSQL (standing for
Not Only SQL—not No SQL as is commonly thought). This technology has arisen as
a result of an increasing need to process “Big Data”, a terribly vague term referring
to data that has at least one of the characteristics of complexity, rapidity of update,
and/or size; these characteristics are often summarized as the three Vs of Variety,
Velocity, and Volume. Here conventional SQL databases are seen to be inadequate
and so new database types are emerging. Triplestores fall within the scope of NOSQL
databases although for some reason the newer NOSQL technologies are seen by some
as competing with Triplestores, RDF, and Linked Data. However the reality is that
the technologies are complementary, and the confusion is more related to a misunder-
standing as to what these technologies are trying to achieve and the niches that each
tries to occupy. Indeed the confusion can be closely related to the misunderstanding
that NOSQL means No SQL, not Not Only SQL. Not Only SQL summarizes the true
situation very well: they are not intended to replace SQL databases but to co-exist
with them. Linked Data fits within this world as a technology very well suited to
handling complexity and data integration, it is not so strong at handling data that is
updated or streamed at very high rates.
One thing that we are very certain of, and this topic would be unnecessary if it
were not true, is that the amount of GI will increase and its scope will broaden.
It will do so as the Linked Data Web grows at an increasing rate, much as the docu-
ment Web did.
We believe there will be a snowball effect in the emergence of commercial Linked
Data. Companies seeing their competitors publishing Linked Data will realize that
they also must have a presence on the Linked Data Web to drive traffic to their sites.
Just as having a Web site is no longer a mere vanity project for companies, so will
linking in to the Linked Data Web become a business essential as a way of dissemi-
nating data or as a fundamental publishing method. While early commercial Linked
Data projects have been motivated by branding or a wish to try out the technology,
primarily based on the subsidy business model, the subscription and pay-as-you-go
models will emerge, probably through a freemium route. This in turn will necessi-
tate development of better methods, it is hoped standards driven, to enable authenti-
cation and payment for Linked Data.
As we have pointed out, software tools to support link discovery and data reuse
(particularly around the areas of evaluation of data quality, trust, provenance, and
licensing) are still relatively immature. We predict that this situation will improve
as tools emerge from university settings and become more robust and commercial-
ized. There is a strong need for more analytical tools, especially with respect to data
mining, and we also see a place for Linked Data Web analytics to track triple usage
and provide publishers with more accurate information about which resources are
