1
Do You Know the Way to SNA?: A Process Model for
Analyzing and Visualizing Social Media Data
Derek L. Hansen*, Dana Rotman*, Elizabeth Bonsignore*, Nataša Milić-Frayling†,
Eduarda Mendes Rodrigues†, Marc Smith
‡
, Ben Shneiderman*, Tony Capone†
*University of Maryland, Human Computer Interaction Lab; †Microsoft Research;
‡
Connected Action

ABSTRACT
Traces of activity left by social media users can shed light
on individual behavior, social relationships, and community
efficacy. Tools and processes to analyze social traces are
essential for enabling practitioners to study and nurture
meaningful and sustainable social interaction. Yet such
tools and processes remain in their infancy. We conducted a
study of 15 graduate students who were learning to apply
Social Network Analysis (SNA) to data from online
communities. Based on close observations of their emergent
practices, we derived the Network Analysis and
Visualization (NAV) process model and identified stages
where intervention from peers, experts, and an SNA tool
were most useful. We show how the NAV model informs
the design of SNA tools and services, education practices,
and support for social media practitioners.
Author Keywords
Social network analysis, visualization, social media,
process model, NodeXL, online communities.
ACM Classification Keywords
H5.m. Information interfaces and presentation (e.g., HCI):
Miscellaneous.
INTRODUCTION
Social media services, such as Facebook, Twitter, Digg,
among others, have enabled new forms of collaboration and
interaction in nearly every imaginable human endeavor.
And we have only begun to realize the potential of
technology-mediated social interaction. Despite numerous
success stories, we must remember the countless failures
due to social and technical factors. How can we support
practitioners in their efforts to cultivate meaningful and
sustainable online interaction?
One promising strategy is to provide tools and concepts that
help practitioners make sense of social media data. There is
precedence to this approach in the development of
sophisticated, yet fairly intuitive website analytics tools
such as Google Analytics [12]. These tools help non-
programmers understand website traffic patterns so they
can make more informed design decisions. We envision an
equivalent set of social analytics tools (e.g., [17, 21]) to
help social media analysts and community administrators
make better decisions based on their in-depth understanding
of social participation and relationships. Social analytics,
which includes Social Network Analysis (SNA), extends
already complex graph analysis metrics and visualizations
with exploratory data analysis approaches, and requires the
engagement of professionals experienced in social
interactions and social media contexts.
To gain acceptance by a broad range of practitioners, tools
that reduce the complexity of data processing are vital.
Eliminating the need to program custom algorithms for
common processing tasks can make SNA more accessible.
Moreover, enabling interactive visual exploration of data
via a variety of layouts can aid in the discovery,
understanding, and presentation of network properties. To
varying degrees, several SNA toolsets such as UCINET,
Pajek, SocialAction, and NodeXL have advanced toward
these goals. However, as with any new practice, success
depends on the common language and best practices that
evolve among practitioners as they apply the tools in
various scenarios and share their experiences and expertise.
We need to understand and capture the processes that
emerge as users explore social interaction to enhance their
power to make sense of and manage interaction patterns.
With that in mind, we conducted a qualitative user study of
graduate students learning to apply SNA concepts and tools
to better understand online communities of their choice. We
make two primary contributions. First, we derive the
Network Analysis and Visualization (NAV) process model
that emerged from the collective experience of students
learning to use SNA metrics and visualizations. We
identified stages within the model where interventions from
peers, experts, and analysis tools are most useful. Second,
we offer recommendations for making SNA tools and
services more accessible to practitioners, especially
novices. These include recommendations for (1) designers
of SNA tools, (2) educators introducing SNA concepts to
online community analysts, and (3) practitioners struggling
to make sense of social media data. We found the fine-level
granularity of the NAV process model invaluable when
developing these recommendations – far more helpful than
more generic sensemaking models, although the NAV
model shared some high-level similarities with them.

RELATED LITERATURE
SNA and its mathematical companion, the graph theory,
have a long and distinguished history of academic
contributions [3, 11, 23]. Recently, many researchers have
used SNA to examine social interaction in computer-
mediated environments, helping to identify unique social
roles [26], social structures [6], and dissemination patterns
[2]. Despite SNA’s success in academic circles and its
appearance in mainstream publications [3] and management
literature [9], it has not been widely used by practitioners.
This is likely to change as more usable SNA tools are
developed and as the historically onerous process of
network data collection is replaced by automatic data
collection from social media sources.
Process models that describe key activities, tasks,
cognitions, and/or feelings have been useful in helping
design novel tools [19] and educational interventions [13].
They are particularly good at identifying moments where
interventions from peers, experts, or computational aids are
most useful. Pirolli & Card [19] call these moments
“leverage points” in their process model of information
analyst’s activities. Their work is part of a larger effort to
characterize the sensemaking process of expert intelligence
analysts [16, 22]. In a different, but related domain,
Kuhlthau [15] developed a process model of information
seeking behavior and identified stages where information
mediators, i.e., educators, can help students the most.
Motivated by the success of these approaches we
investigate the sensemaking process that emerges when
both the SNA concepts and SNA tools are introduced as the
means of data analysis. Our work extends existing literature
on sensemaking models [16, 19, 22] on two fronts: (1) we
observe novices, not experts, and (2) we focus on the social
network analysis and visualization, tasks that have not been
explicitly investigated from the sensemaking perspective.
Considering SNA toolsets, the development of
SocialAction [18] was based on a Systematic Yet Flexible
(SYF) framework that extended successful process models
such as Amazon’s checkout, TurboTax’s income tax
preparation, and the Spotfire Guides for visual analytics.
The SYF framework organized network analysis into 7
steps: (1) overall network metrics (2) node rankings, (3)
edge rankings, (4) node rankings in pairs, e.g., degree vs.
centrality, plotted on a scattergram, etc., (5) edge rankings
in pairs, (6) cohesive subgroups, e.g., finding communities,
and (7) multiplexity, e.g., analyzing comparisons between
different edge types, such as friends vs. enemies. These
steps frame the process that experts – not students or
novices – follow when exploring complex data sets.
Finally, the rapidly growing literature about information
visualization often examines systems that support network
visualization but in-depth user studies such as [18] and ours
are rare. Heer and boyd [14] demonstrated that novices
enjoy browsing data from social networking sites like
Friendster. A survey of 77 researchers, mostly social
scientists, showed a preference towards menu-driven
general purpose packages, such as UCINET and Pajek, over
programmable systems such as JUNG, GUESS, and
Mathematica [1]. However, users expressed significant
frustrations with all the systems due to challenges of
learning complex interfaces [1]. The success of ManyEyes,
a collaborative system for creating and sharing information
visualizations, suggests the desire of many to make
information visualization more accessible [25].
Our detailed user study using an SNA toolset equipped with
a robust set of graph layout options can help characterize
the process novices follow when analyzing network data
with the aid of visualization tools, and offers insights useful
to designers, educators, and community analysts hoping to
broaden the adoption of SNA toolsets and expand a
collaborative SNA knowledge base.
METHODS
Study Setup
We conducted a month-long user study of 15 students in a
graduate course on Computer-mediated Communities of
Practice (CoP).
Teaching Context
The CoP course is an elective, drawing graduate students
from library science and information management. The
purpose of the course is to help students become proficient
community analysts, able to identify and apply appropriate
technologies and social practices to help cultivate
communities. The course includes a weekly classroom
session and a website where students post weekly to a
discussion forum and periodically to individual blogs.
The study took place during a 3-week SNA module that
introduced SNA concepts and the NodeXL SNA tool, and
their application to the data from online communities. The
module occurred 1/3
rd
of the way through the CoP course
and required students to analyze a community they had
chosen to study throughout the semester.
The SNA module included a 2.5 hour, hands-on lab session
that used the Network Analysis with NodeXL: Learning by
Doing tutorial [13]. The tutorial followed a task-based
framework of 9 steps: (1) basic, (2) layout, (3) visual
design, (4) labeling, (5) filtering, (6) grouping, (7) graph
metrics, (8) clustering, and (9) advanced. Course readings
and discussions covered SNA and community metrics,
social roles, and network visualization quality as measured
by NetViz Nirvana guidelines about the network layout
[10]: (1) every node is visible, (2) the degree of every node
can be counted, (3) every edge can be followed from source
to destination, and (4) clusters and outliers are identifiable.
NodeXL SNA Tool
NodeXL is a plug-in for Excel 2007 that exploits a widely
used spreadsheet paradigm to provide a range of basic
network analysis and visualization features [20]. The
NodeXL template is a highly structured workbook with