Information Technology Reference
In-Depth Information
This project aims at a data-driven approach to the problem of transforma-
tive research identification. Specifically, we will focus on the biomedical sciences.
One of the most active and heavily funded areas, the biomedical sciences impact
both public health and economy. Recognizing this need and opportunity, NIH
has established a new oce called Oce of Portfolio Analysis (OPA) in 2011
responsible for applying data-driven approaches to funding prioritization and
resulting impact evaluation. In a pilot study [3], we have developed a prelim-
inary approach to quantifying transformative research with a disruption score
and showed that the method can successfully identify transformative scientific
papers that disrupt established paradigms in Physics and Computer Science. In
this paper, we will show that the disruption score is an accurate early indicator
of a successful transformative research in biomedical sciences as well, regardless
of whether the challenger paradigm is an instant hit or a classic whose con-
tribution is formally recognized with a Nobel Prize decades later. In addition,
this paper will present a model-based approximation algorithm to accelerate the
computation of disruption scores for the first time.
Our approach is novel because it measures disruption in terms of a compe-
tition between information cascades using a recently developed framework to
quantitatively characterize growing information cascades [4]. Using this frame-
work we are able to target transformative challengers of dominant paradigms,
rather than investigate them in isolation, as was done by prior studies. This
allows us to separate influential research that reinforces dominant paradigms
from those that transform them. The field of Bibliometrics in Information and
Library Science focuses on measuring scientific impact of research but rarely
targets transformative research explicitly. Scientific impact is usually measured
by citation counts, and citations-based measures, such as the h-index [5], are
widely used to evaluate the productivity of scientists. Disruptions of citation
cascade growth of well-established, field-defining papers usually represent an
event of “paradigm shift,” “breakthrough,” emergence of a “disruptive idea,”
and a successful “transformative research.” The closest studies to the proposed
work include [6] and [7]. Mazloumian et al. [6] characterize how citations are
boosted when a work is recognized by a Nobel prize. Their goal is to charac-
terize changes in citation patterns after a transformative research work is rec-
ognized, rather than detecting them in advance. Chen [7] describes the use of
a time progressive co-citation network to reveal “intellectual turning point” pa-
pers, which include hubs, landmarks and pivot nodes in the network. Those are
characteristic of highly influential works or works that bridge two important
fields. However, they do not explicitly identify how one dominant paradigm is
challenged by transformative research like the research here.
2 Approach
2.1 Overview
Figure 1 illustrates the idea behind our method. In (a), we show a citation
network, where node 1 and 2 are papers advocating some dominant scientific
 
Search WWH ::




Custom Search