Information Technology Reference
In-Depth Information
taxonomies has two major difficulties: (1) it is very labour intensive, and
thus expensive, and (2) rigid taxonomies tend to lag behind the subjects
that they are trying to categorize, especially in fast-moving fields. In the
generation of folksonomies, the effort is distributed across the users,
which, in addition to spreading costs, also means that categorization
can evolve more quickly. Of course, the downside is a lack of control
over the taxonomy.
Data and text mining
Perhaps the most important leap forward in the discovery process will
come not from the interaction between humans and computers but
from computer-computer interactions in the form of text and data
mining. Allowing computers to trawl and analyse the literature will
reveal latent facts: for example, previously unrecognized relationships
between genes or drug interactions.
Success in this area will require the entire literature to be exposed to
text and data mining, and results may be improved by content
enrichment, the emergence of semantic standards, etc. Exposing the
literature to mining could be achieved under an OA publishing model
(see later), by publishers licensing material, or by their voluntarily
exposing text and data in a machine-readable form for mining.
Challenges of stewardship
Data
There has been a disappointing lack of engagement regarding data
among most stakeholders in the scholarly communication chain, and
this includes academics, research institutions, librarians and publishers.
So-called 'big science' projects - the very large experiments like the
Large Hadron Collider - have data storage and stewardship built into
their core design, and are well catered for. However, the situation for
bench science is very different and much more chaotic. Most publishers
have the facility to store datasets alongside research texts but, with a
few notable exceptions like the Organisation for Economic Co-
operation and Development, they have done little to encourage authors
to submit data and less to link data with the primary research literature.
The result is that information is being lost and the job of the
