TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS
A Case Study
John Haggerty and Sheryllynne Haggerty*
School of Computing, Science & Engineering, University of Salford, Salford, Greater Manchester, M5 4WT, U.K.
*School of History, University of Nottingham, University Park, Nottingham, NG7 2RD, U.K.
Keywords: Social network analysis, Information visualisation, Case study.
Abstract: Social network analysis has received much attention across disciplines. Recently, historians have begun to
complicate their understanding of networks and are increasingly using visualisation to elucidate tangible and
intangible information in their data sets. Current approaches make use of static social network analysis and
whilst they provide a number of tools to explore a social network, they are unable to provide temporal
analysis. This paper presents Matrixify, a practical visual application for the exploratory temporal analysis
of social networks for historians and others not familiar with visual representations of data. This approach
aims to deconstruct the complexity of social networks to provide a temporal analysis to answer a key
problem in historical studies, that of ‘analysis of change over time’. In this way, historians are able to
identify real relationships in their data sets; actors in contact at a particular point in time and shown over
time. The case study presented in this paper demonstrates the applicability of the approach in inter-
disciplinary studies.
1 INTRODUCTION
Social networks and their visual analysis have
received much attention across disciplines. Much of
this interest is due to the explosion of social network
services and the adoption of sociological concepts in
the arts. Therefore, particular disciplines, such as
History, have become increasingly interested in the
results that can be obtained by performing social
network analysis on their data.
A social network is a group of actors that are
defined by their relationships to each other and the
network itself. These relationships are forged
through endogenous and exogenous events affecting
the network. These events may provide or be
supported by qualitative data to elucidate how these
events affect the dynamics of the group, and in turn,
the network’s reactions to such events. Social
networks are complex as they represent both
tangible information (i.e. events within the network,
time of event, etc.) and intangible information (i.e.
relationships between actors, actors’ involvement
within the network, etc.). As Churchill and
Halverson (2005) note, much social network
analysis is relational rather than attribute based. In
addition, networks may be present in many forms,
for example, online communities, business networks
or social contacts. Moreover, an actor will have a
presence in a variety of networks representing
different facets of their social existence.
Many historical studies have presented a benign
view of networks based on familial, ethnic and
religious networks to reduce moral hazard (see for
example, Hamilton, 2005; Mathias, 2000; Walvin,
1997). More recently however, historians have
started to complicate their conceptions of networks
and have begun to look at more formal and civic
arenas especially within an urban environment.
Studies, such as Lipp (2005), Carlos et al (2008) and
Haggerty and Haggerty (2010), have visualised and
measured networks in an attempt to understand their
dynamics. These studies use static network analysis
tools i.e. node-link visualisations. Such applications
focus on quantitative analysis by deriving statistics,
extracting or fitting models and calculating metrics.
However, an important question in historical studies,
that of ‘analysis of change over time’, is not
sufficiently answered using such applications.
Therefore, this discipline has a requirement to
visualise and explore social network data beyond
this static representation.
This paper presents Matrixify, a practical visual
207
Haggerty J. and Haggerty S..
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study.
DOI: 10.5220/0003315802070217
In Proceedings of the International Conference on Imaging Theory and Applications and International Conference on Information Visualization Theory
and Applications (IVAPP-2011), pages 207-217
ISBN: 978-989-8425-46-1
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

application for the exploratory temporal analysis of
social networks for historians. Matrixify aims to
answer the needs of historians and their requirement
to analyse change over time by highlighting ‘real’
relationships; actors in contact at a particular point
in time and shown over a period of time. It provides
an interactive visualisation of time-varying (social)
network data for users to interact with. This
interaction can provide new questions for historians.
Importantly, it identifies macro trends in larger
networks to meet the shortcomings of static social
network analysis tools. It therefore aims to visualise
and provide an interface for exploratory analysis of
both tangible and intangible information, as opposed
to being a ‘measurement’ tool. Matrixify has been
designed from an interdisciplinary perspective,
keeping the literature and discipline problems of
historians, social scientists and computer scientists
in mind. In order to demonstrate the applicability of
the approach, the case study (Membership of the
Company of Merchants Trading to Africa from
Liverpool) presents the questions raised by an
historian using Matrixify. These issues were not
identifiable in a previous study of this historical data
set which used a static social network analysis
approach (Haggerty and Haggerty, forthcoming).
This paper is organised as follows. Section 2
discusses related work in the disciplines of social
science, history and computer science. Section 3
presents an overview of Matrixify. Section 4
discusses historical data sets and their issues for
computer scientists. Section 5 presents a case study
using Matrixify for the exploratory analysis of an
historical social network and the new questions
raised by an historian using this approach. Finally,
we make our conclusions and discuss further work.
2 RELATED WORK
The interaction and dynamics of social networks has
for a long time been of interest in inter-disciplinary
research, and in particular, the social sciences. For
example, Granovetter (1973) highlights the strength
of weak ties in which these relationships provide
new and better opportunities within a network.
Freeman (1978/79) suggests that an understanding
of centrality within social networks may provide
social science researchers with an understanding of
the dynamics of those groups under investigation.
Research conducted by Lawler and Yoon (1996)
focuses on commitment and power in exchange
relationships. This research attempts to predict how
and when people in an exchange become committed
to their relationship.
Therefore, there is a rich
literature from which other disciplines may learn and
incorporate into their own research.
Historians have been using networks as an
analytical tool for some time. This is a worthwhile
exercise because contemporaries also realised the
value of networks, even if they called their contacts
‘friends’ or ‘correspondents’ (Hancock, 2005).
However, it is only recently that this community
have started to use visualisation combined with
statistical analysis. For example, Carlos et al (2008)
used social network visualisation tools to analyse
16
th
Century Jewish merchants’ networks. Haggerty
and Haggerty (2010) visualise an 18
th
Century
Jamaican merchant’s account book to identify
investment groups within Liverpool, UK, and
measured the relationships using graph theory to
highlight key actors. Erikson and Bearman (2006)
highlighted the role of East Indian Company
employees in the extension of business networks in
Asia. The use of information visualisation for
analysis is set to expand within this discipline over
the coming years.
Current tools for social network analysis often
provide static visualisations. Applications such as
Pajek (Vlado, 2010) or SocNetV (Kalamaras, 2010)
visualise network information via the connection of
vertices through arcs and edges. Once produced, the
network and the relationships between actors may be
viewed. These applications provide various tools for
measuring networks. For example, SocNetV
provides a number of measures and layouts for
analysing actors’ power and centrality within the
network based on graph theory. In addition,
weighting may be applied to network edges to
represent strength of ties. The use of weighting (for
example by value or frequency) provides a deeper
analysis (Perer and Schneiderman, 2009). Due to
scalability issues in these visualisations where large
graphs soon become counter-intuitive, other
approaches to network visual representation have
been proposed. For example, Lee et al (2006)
propose TreePlus, a tree layout approach to explore
social networks. These trees can be expanded or
reduced to provide a more intuitive view of larger
networks.
The common issue with static social network
visualisations is that obviously they do not represent
temporal changes in the network, including an
individual actor’s engagement. As discussed above,
this is a key concern for historians. Abello and Korn
(2002) suggest MGV, a system for the analysis of
large data sets. This approach takes underlying data
and represents it in a variety of ways, such as
IVAPP 2011 - International Conference on Information Visualization Theory and Applications
208

location-based graph sketch, needle grid view and
multicomb view. Falkowski et al (2006) suggest an
approach for analysis of subgroup evolution in social
networks. This approach uses a number of views to
facilitate analysis, and displays the network in a
graph, such as those used in static approaches, but
laid out along a temporal plain. Hu and Gong (2010)
present a visualisation of individuals’ spatial-
temporal social networks through three-dimensional
graphs. Belingerio et al (2010) use three-
dimensional graphs combined with hierarchical trees
to identify eras in social networks.
A common issue with these approaches is that
their visualisation of the network may require an
understanding of how they should be read which is
difficult for users not familiar with visual
representations of data. Therefore, Matrixify aims to
simplify the visualisation so that historians may
focus on elucidating and interpreting the information
that they require. This is achieved using a matrix
view of the data. Matrices for analysis of social
networks have been proposed previously. For
example, Snasel et al (2009) propose the use of
Galois lattices for social network analysis. However,
their experiment analyses only a small-scale network
comprising 18 actors over 14 social events and the
visualisation is unable to track single actor
involvement, their role within the network or
provide exploratory tools for further analysis.
GeneaQuilts (Bezerianos et al, 2010) is a system for
exploring large genealogical data sets presenting
information in a sophisticated matrix visualisation.
Alternatively, NodeTrix (Henry et al, 2007)
combines matrices with node-link graphs. However,
both these approaches focus on identifying
communities, actors and roles rather than network
involvement, development and change over time, a
key requirement of historians. Telea and Voinea
(2009) use a matrix-style visualisation on time-series
data. However, this approach focuses on software
maintenance decisions and project management
rather than social network development.
3 MATRIXIFY
This section presents an overview of Matrixify. First,
we discuss the design goals, particularly in relation
to historians’ requirements. We then present the
application and its key features.
A social network comprises a set or sets of actors
that are defined by their relationships. As
Wasserman and Faust (1994) note, it is these
relationships that are the defining feature of a
network. However, it is the events in the network
that create these relationships and affect actor
behaviour. These events ensure that networks are not
static (Lawler and Yoon, 1996). Burt (2004) has also
suggested that bridges or brokers in networks have
the potential to best synthesize information and
thereby have good ideas. Networks are also
instrumental and people engage with them for
different purposes and at different times of their life
cycle or career path (Renzulli et al, 2000).
Therefore, static network analysis visualisations do
not meet historians’ requirements to address the
issue of change over time.
Social networks are complex due to the interplay
between actors, events, content and relationships.
Matrixify attempts to deconstruct this complexity to
present a simplified view of the network, i.e. a
visualisation of dynamic data, for historians to
explore. The design goals of the application have
been derived from discussions with historians and
the findings of previous work by the authors. They
are as follows:
 Visualise temporal events in a network. In this
way, the key issue of change over time can be
addressed.
 Aid exploratory visual analysis of historical
data. Matrixify is designed to produce an
interface for exploratory analysis rather than
presentation or measurement.
 No scripting. Historians with a basic computer
literacy, for example, the ability to use a
spreadsheet to format data, should be able to
use this application.
 Sophisticated network analysis with a simple
interface. Simple data representation aids the
analysis process as users unfamiliar with
visual approaches are able to focus on this
activity rather than trying to interpret the
visualisation.
 Assess individuals or groups of actors and their
pattern of network membership. Historians
often want to highlight the role of individuals
or groups of individuals within a network.
This identification of their actors of interest
aids their assessment of actors’ role(s) in the
wider network.
 Visual clues to identify global network trends.
Use visual representations of data to provide a
global view of the network.
 View network events and their impact on the
network. Historians have a wide range of
cases that would constitute a network event
and the software should be able to adapt to
their needs.
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study
209

Figure 1: Overview of the Matrixify application.
 Ability to export data to common static social
network analysis tools. Whilst the focus of the
software is temporal analysis, static network
analysis provides an alternative view to
explore the networks further.
Figure 1 provides an overview of the Matrixify
application. Matrixify is written in Java and makes
extensive use of Java Swing for the graphical user
interface. Time is placed along the X axis and actor
names down the Y axis. The data is mapped onto the
visualisation and colourised depending on the needs
of the analysis, for example, identifying particular
roles or groups within the network. The application
provides the historian with a number of menus for
both exploratory temporal social network analysis
and export to commonly used static social network
tools such as Pajek (Vlado, 2010) and SocNetV
(Kalamaras, 2010). The key components are
discussed below.
Matrixify File. Data from the historian’s data set is
formatted as a text file containing actor names and a
matrix of numbers delimited by commas (with zero
forming blank spaces in the resulting visualisation).
These numbers represent the colours that will be
visualised by the application. This colourisation
enables the historian to explore known variables
associated with the actors, for example, ethnicity,
family groups, role within the network, etc. to derive
qualitative information from quantitative data.
Data Visualisation. Matrixify reads the file and
places time along the X axis and names down the Y
axis. It also plots the engagement or events in the
network identified in the Matrixify file within the
graph and represents these by colourised dots. It
therefore visually represents an individual actor’s
engagement with the network (reading the data
horizontally) and their network at that particular
point in time (reading the data vertically). In
addition, the colourisation enables an actor’s role
within the network to be easily assessed and
evaluated.
Temporal Social Network Analysis. Whilst the
visualisation provides a rudimentary view of the
whole network and its actors’ engagement, a number
of menus have been added to aid the historian in
their exploration of the data. These menus are as
follows:
 Decade Markers: The addition of decade
markers, grey lines that identify the start of
each decade in the data set, breaks up the total
network view and provides granularity to the
visualisation.
 Change Years: This menu allows the historian
to change the temporal markers to match their
data set, or a particular chronological part of it
 Identify Actors: As discussed above, the
application allows the historian to view actors’
engagement and network change over time by
reading the visualisation both horizontally and
vertically. Therefore, the historian is able to
identify actors by single or multiple years to
view the social network structures for
temporal subsets of the whole data. This
enables them to view the available networks
that actors had engaged in for specific periods
of time and also to compare temporal subsets
with others in the wider network view. This is
achieved by painting a red box positioned
vertically onto the network diagram around
IVAPP 2011 - International Conference on Information Visualization Theory and Applications
210

the temporal subset. However, the historian
may also wish to analyse actors’ roles
developing or changing in the network.
Therefore, they are able to highlight individual
or multiple actors. This is achieved by placing
a red box positioned horizontally by selected
actors across the network diagram.
 List Actors: The historian may wish to view all
the names of actors engaged within the
network during temporal subsets to identify
actors of interest. Data is read vertically and
corresponding names are output to a pop out
window. This can be viewed by year or
multiple years.
 Actor Totals: In some cases, the historian may
wish to view the total number of actors
engaging within the network during temporal
subsets. Therefore, data is again read
vertically to count the total actors engaged
within the network during each year. This is
then output to a pop out window showing
totals by year. Historians can view either a
single year or multiple years.
 Pop-out Directed Graph view: The historian
may wish to view a subset of the data in a
node-link representation as used in static
social network applications. The historian can
select the subset that they wish to view in this
way and Matrixify provides a pop-out window
visualising the network subset. When an actor
is selected in this view, their individual links
within the network are highlighted.
Static Social Network Analysis. Whilst the aim of
Matrixify is to assist historians in their temporal
social network analysis, it is recognised that they
may also wish to conduct static social network
analysis. In this way, they are able to measure the
network using tools commonly associated with these
approaches, such as viewing centrality measures.
Therefore, a menu is available that enables the
historian to convert the data viewed in Matrixify to
common static social network analysis tools, such as
Pajek (Vlado, 2010) and SocNetV (Kalamaras,
2010). This menu produces the files required for
network visualisation in these tools automatically,
thus reducing the need for historians to learn how to
produce these bespoke scripts.
Summary Report Files. Whilst the aim of Matrixify
is to provide an interface to visually explore data, it
is recognised that historians may wish to export
some summary findings for use in other
applications. Therefore, they are able to export actor
totals and actors by single and multiple years to text
files.
Figure 2 illustrates the Matrixify view of part of a
network. An actor’s involvement with the network is
visualised by dots within the visual matrix. Figure 2
demonstrates the data visualisation without any
analysis tools applied apart from colourisation by an
actor’s role, i.e. the data as it is imported into the
application. This data set and colourisation of it will
be discussed later in this paper.
Figure 3 demonstrates the use of the analysis
tools made available by the application. Decade
markers are painted on to the visualisation in grey.
Actors have been highlighted in two ways. First,
individual actors of interest are highlighted by
horizontal red boxes to demonstrate their
engagement within the network. Second, the years
1762 to 1768 have been highlighted by a vertical red
box to identify the actors involved in the network
during this temporal subset.
Figure 4 demonstrates the pop-out window
containing the directed graph view of data subsets.
As illustrated, more than one pop-out window
containing different subsets can be viewed at the
same time facilitating comparison between graphs,
in this case the years 1750-1753 (background) and
1760-1763 (foreground), both subsets of the data
presented in section 5. The graph view shows the
number of actors in the network subset visualised
and the number of relationships between them. As
demonstrated, an actor, once selected, changes
colour to red and their direct links with other actors
are also highlighted in red. In this way, the historian
may explore the connections between actors of
interest within the graph.
This section has presented an overview of
Matrixify, a temporal social network analysis tool
designed for historians and their data. In particular,
this tool visualises data for historians to explore. In
the next two sections, we discuss historical data sets
and present details of our case study.
4 HISTORICAL DATA SETS
It is worth briefly discussing historical data sets to
an audience outside this discipline. Historians rely
on both quantitative and qualitative data for their
studies. Matrixify attempts to provide qualitative
information through quantitative data sources,
especially useful where qualitative data does not
exist. It achieves this by visualising network
information, for example, actor relationships or
engagement within the network over time. In
addition, it identifies network trends in the whole
network over time. Current static social network
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study
211

Figure 2: Matrixify raw data visualisation.
Figure 3: Matrixify using analytical tools.
IVAPP 2011 - International Conference on Information Visualization Theory and Applications
212

Figure 4: Matrixify directed graph pop-out window.
analysis tools are unable to meet these requirements
With a few exceptions, such as oral, art or media
history projects, most of the data that historians
analyse is from contemporary written or printed
documents. These documents are stored in archives,
and the period of history under study will determine
the amount of data available. For example, those
studying modern history may have a variety of
primary sources and many in easy-to-read, printed
form. Those studying medieval history will have far
fewer documents, mostly from religious sources and
often written in Latin. Thus, many historians collect
their data by transcription in archives or through
photographs for later transcription or reference.
Historians often use common software such as
databases or spreadsheets to store their data. These
applications allow easy export of existing data to
text format to be visualised in Matrixify. However,
there is no standardisation in the way data is
collected and it is often down to the historian as to
what is collected and in which format. Moreover,
data sets vary in size, from a few to hundreds of
actors. The data set used for the case study in the
next section represents a large data set for historians.
This leads to a number of issues that must be
accounted for by those outside this discipline. First,
data is either extant or not. Historical records have
been lost through time and historians can only work
with the data that has survived. Therefore, they
focus on the data itself rather than the algorithms for
analysing that data. Second, historical data was
created for a particular purpose at a particular point
in time in order to meet the social or political
requirements of the day. It therefore has a variety of
inherent biases. Furthermore, many data sets that
may have been useful to historians were therefore
never collected at all. Third, due to the extant data
problem, data is finite, therefore mitigating issues of
scalability of visualisation approaches that would be
encountered in studies of modern social networks
such as those identified through electronic means. At
the same time, actors cannot be interrogated in order
to compensate for lacking data. Finally, data
collection is error prone due to a variety of reasons,
including: unfamiliar and/or difficult to read
handwriting which must be interpreted and collected
manually; errors in the translation of languages;
printed documents that are unsuitable for scanning;
manual errors in transcription or database entry; a
lack of ability to check for anomalies in the data, for
example, in our case study family members with the
same name using (or losing) junior or senior for
identification of an individual.
5 CASE STUDY
This section presents a case study using Matrixify
for visualisation and analysis of social networks
identified in historical data sets. In particular, the
questions and issues raised later in this section are
posited by an historian applying the Matrixify
approach to data they have collected as part of their
research into eighteenth-century business networks.
These new questions are of real significance to this
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study
213

Figure 5: Static social network analysis of the African Company data in Pajek.
community.
The aim of using temporal analysis is to visually
explore the data in order to provide a more nuanced
and sophisticated view of the network and identify
change over time. It does not aim to answer
questions; but to raise questions around the data not
evident in other forms. These questions can then, for
example, guide the historian to relevant sources of
further study or challenge existing hypotheses and
theories.
The data we use in this case study is a subset
from a larger study of metropolitan business
networks in Liverpool between 1750 and 1810
(Haggerty and Haggerty, forthcoming). This study
uses static social network analysis tools and
techniques to visualise and measure the relationships
between actors and networks identified by trade,
cultural, political and social institutional
membership during this period. This approach
highlighted a number of issues which provides the
motivation for the development of Matrixify.
Moreover, the questions posited by the historian by
using the Matrixify approach were not identifiable
when using static social network analysis tools in
previous work (Haggerty and Haggerty, 2010;
Haggerty and Haggerty, forthcoming).
The data that forms the case study is membership
of the Company of African Merchants Trading from
Liverpool, hereafter referred to as the African
Company (Company, 1750-1810). Members’ names
were collected from the original historical records
for the period and their attendance at meetings was
recorded in a spreadsheet. This spreadsheet therefore
represents those actors engaging with that network
on an annual basis.
Figure 5 demonstrates the limitations of static
network visualisation for the analysis of network
membership over time, a key concern for historians.
The network is laid out using Kamada-Kawai fixing
first and last actors. As can be seen, whilst the
network is represented and key actors highlighted, it
is difficult to visualise how actors engaged in the
network, to what extent they engaged and how this
changed over time. In addition, network trends, such
as changes to the overall network structure or
individual actor engagement over time cannot be
ascertained.
Figure 6 visualises the same data using Matrixify.
The years are placed on the X axis and actors that
engaged with the network during this time are
represented on the Y axis by name. Their
involvement and role in the network is indicated by
coloured dots. Thus, individual engagement with the
network is represented horizontally and the networks
that were active on a yearly basis are viewed
vertically. Decade markers are included to aid
reference later in this section.
The African Company was a formal trade
association in Liverpool with links to similar
societies in London and Bristol. Set up in 1750, it
was run by merchants, and represented one of the
important trades of the town, the trade to Africa, and
in particular the Atlantic slave trade. Whilst it was
dominated by slave traders, membership was not
exclusive to just those in this trade. Figure 6
illustrates the membership of this trade association.
IVAPP 2011 - International Conference on Information Visualization Theory and Applications
214

Figure 6: Temporal social network analysis of the African
Company data in Matrixify.
The remainder of this section highlights new
general questions thrown up by the historian using
temporal network analysis, and gives specific
examples pertinent to the case study. That is, those
issues particular to business networks identified via
formal institutions in Liverpool during the period
1750-1810 that are not identified through static
social network analysis.
First, why were some actors active for only a
short time and others active for decades? Do some
not find a particular institution useful for
networking, and if so, why? Do they network
elsewhere? Conversely, do the long-term actors
dominate the network, and do they constitute a
clique by family, trade or other interests? Is this a
bad thing for this network? Matrixify highlights that
in this particular case study the network is
dominated by a group of families – the Cases, the
Earles, the Heywoods and the Tarletons. These are
all dominant families in the Liverpool
historiography of this period.
Second, why is the network more dense in
particular periods? In this case, the 1770s and 1780s
and to a lesser extent the 1790s? Why does this
change significantly in the mid-1790s? Are
exogenous or endogenous events driving this? In this
case we would need to look at the dislocation and
problems caused by, and reactions to, the American
War of Independence (1776-1783), credit crises
(1772 and 1793), the movement for the abolition of
the Atlantic slave trade (especially during the 1780s)
and the outbreak of the French Wars (1793).
Third, what are the causes for the relative lack of
actor involvement in certain periods? In the case
study this occurs in the 1750s, 1760s and 1800s.
Why are these periods particularly lacking in long-
term actor involvement? Are actors using other
formal and/or informal networks? In terms of this
case study, the lack of actors in the early period
raises particular cause for comment for historians as
this was when Liverpool rose to prominence in the
slave trade, a key concern of this particular network.
At the other end of the temporal scale, there is a lack
of engagement during the period up to the abolition
of the slave trade in 1807. Furthermore, in contrast
to the 1770s and 1780s, the Seven Years’ War
(1756-1763) and the end of the French Wars do not
appear to be exogenous events that precipitate more
actor engagement.
Fourth, who is chosen to represent the network at
remote locations and why? (London-based members
are represented by red dots.) In this case the position
is often filled by family members (for example,
Henry Blundell, James Crosbie, and Richard
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study
215

Gildart). Therefore, were familial connections
deemed more trustworthy over non-familial, and
perhaps more influential, contacts? Were these
strategies successful? And was one more so than the
other?
Fifth, how does shifting actor engagement
impact on access to information, capital and the
ability to react to exogenous events? In this case,
how did the finance and networks of slave trade
networks change over time and did this affect their
strategy and ability to react to credit crises,
dislocation caused by war and abolition of the slave
trade?
These questions highlighted by the Matrixify
visualisation point to a need to conduct research in
further areas or themes and particular sources. This
is particularly important where there is a lack of
qualitative data about a network itself. In this case
the historian would want to conduct further research
on: the histories of the key families and their
networks (through secondary literature, family and
business papers); the changing responses to
exogenous conditions (through petitions to
Parliament, the records of similar associations in
London, House of Commons Sessional Papers and
Parliamentary Papers); the business networks of the
members (through business letters and accounts in
Liverpool, London and elsewhere); other
institutional membership (other clubs in Liverpool,
Town Council, etc.).
Matrixify therefore provides an exploratory tool
which highlights actor engagement and key trends in
a network. The questions raised by the historian
using such an analysis also point to avenues for
further research which may provide answers to these
questions. This is especially useful where qualitative
data is lacking for a particular network; a common
issue for historians.
6 CONCLUSIONS & FURTHER
WORK
Social network analysis has received much attention
across disciplines. The visualisation of networks
comprising both tangible and intangible information
has enabled researchers to understand how actors
use their relationships in their social lives. Recently,
historians have started to complicate their
understanding of networks in order to analyse their
data sets. This paper has presented Matrixify, a
temporal social network visualisation approach
designed to provide historians with a nuanced and
sophisticated view of their networks over time and
by individual actor. It deconstructs the complexity of
the network to present a simplified view for
historians to explore. In addition, it provides
information and raises questions that would not be
seen when using static social network analysis.
Further work aims to enhance the exploratory
tools available in Matrixify. In particular, further
statistical tools and layouts, such as measures of
centrality, will be included in both the temporal and
directed graph visualisations. In addition, the
questions raised by the case study are currently
being explored by historians to form part of a wider
project in the analysis of eighteenth-century business
networks. Matrixify has demonstrated that a simple
social network view can provide a wealth of
information for the historian to enhance their
understanding of the actors identified in their data
sets.
REFERENCES
Abello, J., Korn, J., 2002. MGV: A System for Visualizing
Massive Multigraphs. In IEEE Transactions on
Visualization and Computer Graphics 8(1) pp. 21-38.
Bezerianos, A., Dragicevic, P., Fekete, J.-D., Juhee Bae,
Watson, B., 2010. In IEEE Transactions on
Visualization and Computer Graphics, 16(6) pp. 1073-
1081.
Burt, R. D., 2004. Structural Holes and Good Ideas. In
American Journal of Sociology, 10(2) pp. 349-399.
Carlos, A. M., Maguire, K., Neal, L., 2008. ‘A Knavish
People…’: London Jewry and the Stock Market
During the South Sea Bubble. In Business History
50(6) pp. 728-748.
Churchill, E. F., Halverson, C. A., 2005. Social Networks
and Social Networking. In IEEE Internet Computing
Sep/Oct 2005 pp. 14-19.
Company, 1750-1810. Minute Book of the African
Merchants Trading from Liverpool, unpublished,
Liverpool Record Office 352/MD1.
Erikson, E., Bearman, P., 2006. Malfeasance and the
Foundations for Global Trade: The Structure of
English Trade in the East Indies, 1601-1833. In
American Journal of Sociology 112(1) pp. 195-230.
Falkowski, T., Bartelheimer, J., Spiliopoulou, M., 2006.
Mining and Visualizing the Evolution of Subgroups in
Social Networks. In Proceedings of the International
Conference on Web Intelligence, Hong Kong, pp. 52-
58.
Freeman, L. C., 1978/79. Centrality in Social Networks. In
Social Networks 1 pp. 215-239.
Granovetter, M. S., 1973. The Strength of Weak Ties. In
American Journal of Sociology 78(6) pp. 1360-1380.
Haggerty, J., Haggerty, S., 2010. Visual Analytics of an
Eighteenth-Century Business Network. In Enterprise
and Society 11(1) pp. 1-25.
IVAPP 2011 - International Conference on Information Visualization Theory and Applications
216

Haggerty, J., Haggerty, S., forthcoming. Life Cycle of a
Metropolitan Business Network: Liverpool 1750-
1810. To appear in Explorations in Economic History.
Hamilton, D. J., 2005. Scotland, the Caribbean and the
Atlantic World, 1750-1820, Manchester University
Press, Manchester.
Hancock, D., 2005. The Trouble with Networks:
Managing the Scots’ Madeira Trade. In Business
History Review 79 pp. 467-491.
Henry, N., Fekete, J.-D., McGuffin, M. J., 2007.
NodeTrix: A Hybrid Visualization of Social Networks.
In IEEE Transactions on Visualization and Computer
Graphics, 13(6) pp. 1302-1309.
Hu, B., Gong, J., 2010. Modeling Individual-Based Social
Network with Spatial-Temporal Information. In
Proceedings of the International Conference on
Management and Service Science, Wuhan, China, pp.
1-4.
Kalamaras, D. B., SocNetV. http://socnetv.sourceforge.
net, accessed October 2010.
Lawler, E. J., Yoon, J., 1996. Commitment in Exchange
Relations: Test of a Theory of Relational Cohesion. In
American Sociological Review 61 pp. 89-108.
Lee, B., Parr, C. S., Plaisant, C., Bederson, B. B., Veksler,
V. D., Gray, W. D., Kotfila, C., 2006. TreePlus:
Interactive Exploration of Networks with Enhanced
Tree Layouts. In IEEE Transactions on Visualization
and Computer Graphics 12(6) pp. 1414-1426.
Lipp, C., 2005, Kinship Networks, Local Government, and
Elections in a Town in Southwest Germany, 1800-
1850. In Journal of Family History, 30, pp. 347-365.
Mathias, P., 2000. Risk, Credit and Kinship in Early
Modern Enterprise. In McCusker, J.J. and Morgan, K.
(eds.) The Early Modern Atlantic Economy,
Cambridge University Press, Cambridge.
Perer, A., Schneiderman, B., 2009. Integrating Statistics
and Visualization for Exploratory Power: From Long-
Term Case Studies to Design Guidelines. In IEEE
Computer Graphics and Applications May/Jun pp. 39-
51.
Renzulli, L. A., Aldrich, H., Moody, J., 2000. Family
Matters: Gender, Networks and Entrepreneurial
Outcomes. In Social Forces 79(2) pp. 523-546.
Snasel, V., Horak, Z., Kocibova, J., Abraham, A., 2009.
Reducing Social Network Dimensions Using Matrix
Factorization Methods. In Proceedings of Advances in
Social Network Analysis and Mining, Athens, Greece,
pp. 348-351.
Telea, A., Voinea, L., 2009. Case Study: Visual Analytics
in Software Product Assessments. In Proceedings of
the 5
th
IEEE International Workshop on Visualizing
Software for Understanding and Analysis, Edmonton,
Canada, pp. 65-72.
Vlado, A., Pajek. http://vlado.fmf.uni-lj.si/networks/pajek,
accessed October 2010.
Walvin, J., 1997. The Quakers: Money and Morals, John
Murray, London.
Wasserman, S., Faust, K., 1994. Social Network Analysis:
Methods and Applications, Cambridge University
Press, Cambridge.
TEMPORAL SOCIAL NETWORK ANALYSIS FOR HISTORIANS - A Case Study
217