Content uploaded by Edward Melcer
Author content
All content in this area was uploaded by Edward Melcer on Jul 13, 2020
Content may be subject to copyright.
COVER SHEET
Title: A Co-Word Analysis of the Structural Health Monitoring Field
Authors: Kaitlyn Kliewer
Edward Melcer
Branko Glisic
ABSTRACT
This research applies a bibliometric analysis to the structural health monitoring
(SHM) research field over a 15-year period, from 2003 to 2017. The goal of this research
is to provide a broader analysis of the field as a whole through a data-driven approach
with a bibliometric analysis of SHM literature. The aim of this work is to understand
relationships between research themes and communities within the SHM field, and how
this landscape has evolved over time. By employing graph theory and social network
analysis methods, this study uses a data-driven analysis using a co-word and co-venue
analysis based on publication data from over 21,000 conference and journal papers in
SHM. Strategic diagrams and network analysis were used to map and visualize the
academic landscape of SHM. Additionally, a co-venue analysis was performed to
determine the conference and journal research communities within the field at each time
segment. The analysis was conducted for three 5-year time periods providing insight on
how the communities within the SHM research field have evolved.
INTRODUCTION
With the improvement of technology, data management, and instrumentation
methods, the structural health monitoring (SHM) research field has grown considerably,
particularly over the past 15 years. Currently, SHM is a broad interdisciplinary field,
encompassing research on sensors, civil, mechanical, and industrial structures such as
civil infrastructure, lifelines, and aircraft, data acquisition, processing and analysis, data
management, and diagnostics. With such a far-reaching research community, it can be
challenging to understand where the field currently stands, where your research fits into
the field, and where there are existing gaps in current research. There have been
numerous literature reviews on SHM research and the communities within SHM that
help provide an overview of field. However, literature reviews are limited in nature and
_____________
Kaitlyn Kliewer, LERA Consulting Structural Engineers, 40 Wall Street, New York, NY 10005,
U.S.A. Email: kaitlynkliewer@gmail.com
Edward Melcer, University of California Santa Cruz, Department of Computational Media, Santa
Clara, CA 95054, U.S.A. Email: eddie.melcer@ucsc.edu
Branko Glisic, Princeton University, Department of Civil and Environmental Engineering,
Princeton, NJ 08544, U.S.A. Email: bglisic@princeton.edu
typically cover no more than 100 papers. Thus, they tend to be more focused on a sub-
community within SHM, as opposed to an analysis of the interrelations among research
themes in the field. The aim of this research is to provide a better understanding of the
relationships between these research communities in SHM and provide an overview of
how this landscape has evolved over time.
Bibliometric data was collected and analyzed from a total of 21,251 conference and
journal papers from the period from 2003 to 2017. This data was broken into three five-
year time segments to analyze the evolution of the field. A co-word analysis was used
to determine the core research themes for each time period and the results were
visualized using strategic diagrams and keyword network maps.
METHODOLOGY
A co-word network analysis is a method that has been applied extensively in a wide
range of research fields, ranging from computer science [1] to consumer behavior [2].
However, to the authors current knowledge, these methods have yet to be applied to the
SHM field. A co-word analysis determines the themes and linkages between those
themes using the co-occurrence of pairs of keywords [3,4]. This method requires no a
priori knowledge of the themes within the research network. The primary assumption
and thus, when two keywords occur on the same paper, there is a link between those
two topics. From the co-word analysis, a weighted co-word network graph is created
and the research communities in the network can be
community detection algorithm [5].
RESULTS
Data from over 21,000 research papers were collected using the Scopus database.
From each paper, the authors, title, publication year, publication venue and keywords
were collected. The total number of conference and journal papers collected for each
year is shown in figure 1. It was determined this was a sufficiently large sample of the
literature in the field to provide a representative sample of the research in the field and
this was confirmed by a completeness verification analysis performed on the data set.
Form those papers, over 40,487 unique keywords were determined. These keywords for
papers from 2003 to 2017 are illustrated in a keyword cloud visualization shown in
figure 2 where the size of the keywords is proportional to the keywords frequency. Over
the 15-year
papers.
A weighted co-word network graph was constructed for each time period where the
nodes in the network graph represent keywords that are linked together by an edge
weight. The edge weight is based on how often the two keywords occur on the same
paper together. When the keywords for a community are closely spaced to one another,
Figure 1. Combined number of conference and journal papers from which bibliographic data was
collected for this analysis by year.
Figure 2. Keyword cloud visualizing for SHM papers from 2003 to 2017. The font size of each keyword
is proportional to its occurrence frequency.
the research theme has high density, indicating the community high internal cohesion.
Additionally, when a keyword or a research cluster appears closer to the center of the
network graph, the keyword or cluster is more central and core to the SHM field. An
example of a network graph for the period from 2007 to 2013 is shown in figure 3. Each
color in the network graph illustrates a different research community that was found
using a community detection algorithm. Figure 3 illustrates a partial selection of the
network from this time due to space constraints, however full network graphs were
created for each of the three time periods. From those networks, the major research
themes were determined for each time period. A total of 11 themes for 2003-2007, 15
for 2008 to 2012 and 15 for 2013-2017 were found.
Overall the network graphs provided a visualization of the growth of the community
over 15-year analysis period. It was found that as the field matured, the communities
became broader but more connected to other topics. This potentially indicates an
increase in interdisciplinary research spanning across the communities. The research
0
500
1000
1500
2000
2500
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
Total of Papers
Year
themes relating to sensor networks remained central for all three time periods. The
analysis also showed the emergence of communities, such as wireless sensor networks,
and the declining of other research communities, with declining centrality and density.
Figure 3. Partial keyword network map 2013-2017.
CONCLUSION
The aim of this research was to provide an overview of the communities within the
SHM field and the evolution of those communities over the past 15 years. A data-driven
approach was used to determine the major themes and communities for the three 5-year
time periods. The results illustrated the notable growth the field has experienced over
the past 15 years with an increase in centrality. Additionally, the results confirmed the
increase of interdisciplinary research communities and illustrated the interactions
between these research clusters.
REFERENCES
1. Melcer, E., Truong-Huy Dinh Nguyen, Zhengxing Chen, Alessandro Canossa, Magy Seif El-Nasr,
-
Network, 17:20, 2015.
2. Munoz-Leiva, F., Maria Isabel Viedma-del Jesus, Juan Sanchez-Fernandez, and Antonio Gabriel
Lopez- -word analysis and bibliometric maps for detecting the most
Quantity, 46(4):1077{1095, 2012}.
3. e and technology: Sociology
4.
ics,
27(2):119{143, 1993}.
5. Blondel, V., Jean-
2008(10):P10008, 2008.
