Citation Networks: A New Humanities Tool?

This paper proposes to use citation networks as a tool for
mapping out links among disciplines and research areas
in & around humanities. By collecting papers from main
journals in diverse areas like cognitive science, art history and
psychoanalysis, this study aims to build a small citation network
that will be visualized and published as a 3d web page. The
end-map will be the backbone of a dissertation, which tries to
analyze the influence of various movements and disciplines
onto the art historical canon. Used as such, a citation network
can become more than a frozen map that has more than a
one-time use value.
Access to online electronic databases and tools such as
Google scholar have led to a significant improvement in
the discovery of secondary literature. However, organizing the
vast amount of bibliographic information from various
discipline-specific databases continues to be an impediment to
truly interdisciplinary work. My project attempts to construct
virtual maps of electronic databases or digital libraries capable
of providing scholars with significant links between disciplinary
relations, interdisciplinary research areas, or tendencies,
approaches and methodologies inside a single discipline.
Moreover, as I will show, if the third dimension, namely history,
is added to these maps, the transformation of the disciplines,
the merging of research areas, and the changes in the taxonomic
structure of the academy will be revealed to the expert eye. As
an example of how such a virtual time-map could become a
valuable analytic tool during the research process, I will
construct such a tool and, using my dissertation as a test case,
demonstrate how it can be applied to map intersections between
art historiography, psychoanalysis and cognitive science.
As a PhD candidate in UCLA's Art History program, my
dissertation traces the changes of the critical discourse from the 1970s, when the so-called "New Art History" clashed with
traditional art historians, and gave rise to a whole new
approach—one that has now become known as Visual Cultural
Studies. However, in such a broad context, one needs to handle
a deluge of texts and interrelations. A simple timeline, a linear
outline constituted of chapters and subchapters is not enough
to depict the map of overlapping layers, concepts and relations
between opposite but—in this case still—allied methodologies.
In order to render all these visible, I would like to create a
multidimensional space of such relations. To that end I propose
to collect significant papers, extract citation information using
various text-analysis programs and visualize the end results as
a citation network that runs along three different trajectories,
namely the history of psychoanalytical and cognitive scientific
methodologies and their impact on the evolution of New Art
History into Visual Cultural Studies.
History:
Arelatively new research venue, called ‘scientometrics’ or
‘bibliometrics’, specializes in creating such maps for
delineating growth, relations and interactions in scientific
fields.1 Bibliometrics uses text analysis to extract citation data
from papers and makes use of this data as a way of evaluating
scientific publications. To obtain information from scientific
papers and using he results in mapping out scientific relations
has a long history. As early as 1964 Garfield and his colleagues
suggested using citation data to evaluate the development of
science.2 From 80’s on, the research in this area accelerated
with the advancement of computers and various combinations
of statistical methods used to extract and evaluate information
such as citations, co- citations with reference of various
bibliometric data. The end-results are usually rendered as
so-called ‘citation networks’ which are a variation of social
networks.3 Now it is a common practice to evaluate a scholar
or a journal according to how many times it/he/she is cited.
Moreover, there is so much research done using citation
networks as a methodology to analyze the disciplines, many
scholars are now questioning the efficacy of this approach.4
Among the ample publications in this area, two general
approaches distinguish themselves: the citation networks are
either built to support an idea or to enhance the way in which
such networks are composed. In the first instance, a search is
done to filter out unnecessary papers. The maps generated in
this way are limited with a scope, and generally give an
overview of the research topic. These maps are not created from
a relational database and are not published for further use. That
means they are not applicable to other research questions and
the enormous work put into collecting papers and preparing
them for network analysis is done on a case-by-case basis
because access to the databases is restricted and therefore the
datasets themselves cannot be made available. In the second
instance papers are extracted from the electronic database
without any filtering. Instead a time limit is imposed. These
types of publications tend to focus more on the technical details
and explore the mathematical substructure of social networks.
Usually open-source databases are preferred, since the main
idea is to test the application’s performance on huge datasets.
Papers about such studies do not interpret the resulting map
and instead detail the mathematical innovation of the
applications.
Despite the popularity of this approach in the sciences, I have
yet to find a paper that uses Humanities databases. Even the
most comprehensive citation network, a data set that
encompasses "7,121 journals covering over 1 million documents
in the combined Science Citation and Social Science Citation
Indexes” does not delve into the Arts and Humanities Citation
Index.5 Rather than simply applying the same techniques to
Humanities materials, a fresh approach, one that is not only
suits to humanities scholarship, but addresses some of the issues
raised in the scientific communities seems desirable.6 Although
my long-term goal is to create a dynamic citation network that
can become a part of an ongoing research project in digital
humanities, in this paper I will focus on creating a static citation
network using open-source tools. My research plan to achieve
this aim is as follows:
1. Collecting papers: I have already collected around 2000
papers, mostly from prominent journals in Art History (such
as Art Bulletin, October, Art Journal and Leonardo), and
in Cognitive Science (Trends in Cognitive Sciences). Beside
these resources I would like to include the classical texts in
psychoanalysis; which can be found in electronic format in
the Psychoanalytic Electronic Publishing database. The
main criterion used for selecting pertinent papers is to use
keywords that are relevant for the research topic. For
example I used keywords such as "Freud", "psychoanalysis",
"aesthetic", "artist" while searching in a cognitive science
journal whereas I chose keywords like ‘cognition",
"cognitive science", "vision", "artist" etc. for the search in
the domain of psychoanalysis.
2. Preprocessing the database for text analysis & extraction
the needed information: The acquired corpus is preprocessed
for text mining and analysis. Preparing a list of keywords
and bibliometric data (author name, date, journal name,
title, etc.) will be enough for the preprocessing stage. We
will experiment with different text- mining and text analysis
programs and report on those that work well at this stage
of data preparation.
3. Construction of social networks: A social network is a graph
representation of social relations. Graphs are the most
popular and widely researched data structure for
representing and processing relational data. In a graph, each
node represents one entity (a person in a social network; a
researcher or a work in a citation network) and the edges (or arcs, if they are directed) of the graph represent some
relation. One can also indicate the strength of the relation
by associating weights with the edges of the graph. Then,
by using tools like Pajek7, the graph nodes can be placed
in a 3D space in such a way to minimize an energy term.
Thus, the nodes that are close to each other semantically
(through the interpretation of graph edges) are placed in
proximity, even when they don't have actual links. On a
social network, clusters and cliques can be identified,
indirect relations can be uncovered, and relevance judgments
can be made based on quantitative or qualitative measures.
Even the location of the nodes (center vs. periphery) can
be informative for a person thoroughly acquainted with the
represented structure.
The use of such a graph tool is simple; nodes and arcs are read
from a file, and the graph visualization is accomplished with a
few commands. For a citation or semantic network, text mining
tools can be employed to derive the entities in relation
automatically. Once such a network is built, Pajek can import
the graph in 3D file formats like 3xd and VRML; both are now
becoming standards for internet publication in 3D.8
Conclusion
The proposed tool serves not only as a bibliographic aid,
but will become the main framework for my dissertation.
By incorporating the 3D virtual map into my dissertation, I
hope to demonstrate a new form of digital scholarship--one that
springs from the new possibilities that digital technologies
affords scholars in the humanities whose work is inherently,
and exuberantly, interdisciplinary.
1. To read more on the history of scientometrics see Katy Börner,
Jeegar T. Marus, and Robert L. Goldstone. "The Simultaneous
Evolution Of Author And Paper Networks," PNAS 101 (suppl.1)
(2004): 5266-5273.
2. All these search engines providing citation index information are
products of Thomson ISI. The original foundation was called
simply Institute for Scientific Information. Garfield launched it,
again in 1964, see ibid.
3. Doreian Patrick, "A Measure Of Standing Of Journals In Stratified
Networks," Journal of the American Society for lnformation
Science 8.5-6 (1985): 341-363.
4. See R. E. Rice et al., "Journal-To-Journal Citation Data,"
Scientometrics Vol. 15.3-4 (1989): 257-282; Lindsey D, "Using
Citation Counts As A Measure Of Quality In Science Measuring
What's Measurable Rather Than What's Valid," Scientometrics
15.3-4 (1989): 189-203; Nederhof, A. J., and Zwaan, R. A.
"Quality Judgments of Journals as Indicators of Research
Performance in the Humanities and the Social and Behavioral
Sciences," Journal of the American Society for Information
Science 42.5 (1991): 332– 40; Nederhof Anton J., "Bibliometric
Monitoring Of Research Performance In The Social Sciences And
The Humanities: A Review," Scientometrics 66.1 (2006): 81-100;
Leydesdorff Loet, "Can Scientific Journals Be Classified in Terms
of Aggregated Journal-Journal Citation Relations Using the Journal
Citation Reports?," Journal Of The American Society For
Information Science And Technology (March 2006): 601-614.
5. Boyack, Kevin W., Klavans Richard, Börner Katy," Mapping The
Backbone Of Science," Scientometrics 64.3 (2005): 351.374
6. Even though the end results of the “citation networks” give a
scholar a good overview, they are still not integrated into the
academic research facilities like Web of Science, Science Citation
Index, Social Citation Index or Arts & Humanities Citation Index
etc. The main reason is that once a citation network is derived, it
becomes a static entity; it covers a limited time and scope. Thus
citation networks, by their very definitions and aims, fail to keep
up with new publications.
7. You can find more information about Pajek, its history and
application areas at <http://vlado.fmf.unilj.
si/pub/networks/pajek/default.htm>, last
accessed 2006-11-15.
8. Please check <http://www.w3.org/> to see the latest
standards of World Wide Web in relation to 3d publishing, last
accessed 2006-11-15.