Author connectivity within disciplines: how big is your world?
Acuna D, Novoa E. Author connectivity within disciplines: how big is your world?. Oral presentation at 2017 SciTS Conference. Clearwater Beach, FL. Jun 13, 2017. Insights Through Network Analysis. Online at: http://www.scienceofteamscience.org/2017-agenda.
Collaboration in science is important for improving the
productivity, quality, and reproducibility of research.
Often, new research projects start when scientists
make connections among their advisors, mentors,
and co-authors. Understanding the connectivity (e.g., distance) between two scientists is therefore important
for starting new collaborative projects. However, we
do not know how well scientists are connected within
disciplines because access to datasets have been
limited. Recent releases of datasets open the door to
explore these answers.
Previous researchers have examined how disciplines
differ in collaboration patterns. Newman (2001, 2004) found that scientific collaborations in Biomedical
Research, Physics, and Computer Science exhibit
a “small world” pattern, where two scientists are
connected by a short path. Finardi and Buratti (2016)
found strong patterns of collaboration across BRICS
countries and Coccia and Wang (2016) found stronger
collaborations between applied and basic sciences in
the course of decades. However, these studies have
been relatively small in scale.
The purpose of this work is to understand author
connectivity differences across disciplines. Our
hypothesis is that scientists are closely connected but
with significant differences across fields. To answer
this question, we analyzed the Microsoft Academic
Graph (Sinha et al., 2015), matched to the Web of
Science discipline classification, containing 16.9
million authors and 25 million articles. We analyzed
19 fields (Fig. 1 A and B) and computed the most
frequent field of publication per author. We found
that the Multidisciplinary field, with a small share of
only 2.2% of the authors (Fig. 1A), has the shortest
average distance (3.41, Fig 1B) whereas Engineering
has the highest (6.75). Similar to previous studies
(Newman, 2001), we confirm that there is a significant
correlation between the log of the number of authors
and the log of the number of papers with the average
distance within fields (r(18) = 0.52, p = 0.02, and r(18) =
0.52, p = 0.02, respectively). Our results coincide with
previous findings that the sciences have relatively high
minimum distance (compare Agricultural and Biological
Sciences = 6.45 with Physics and Astronomy = 5.2).
We will explore ideas for future work where we can
create systems to automatically suggest collaborators
based on fields of study. These results shed light on
how teams of scientists may be formed based on
connections among them.
Type of Publication:
scits 2017 conference, presentation, connectivity, connections, world, collaborations, author connectivity
Addresses these goal(s):
- Learn about the field of team science: history, theory and concepts
- Conduct research on/evaluate team science
Resource created by Jane Hwang on 10/3/2017 5:29:40 PM.