Unveiling Facebook: A Measurement Study of Social
Network Based Applications
Atif Nazir, Saqib Raza, Chen-Nee Chuah
University of California, Davis
{anazir, sraza, chuah}@ucdavis.edu
ABSTRACT
Online social networking sites such as Facebook and MyS-
pace have become increasingly popular, with close to 500
million users as of August 2008. The introduction of the
Facebook Developer Platform and OpenSocial allows third-
party developers to launch their own applications for the
existing massive user base. The viral growth of these social
applications can potentially influence how content is pro-
duced and consumed in the future Internet.
To gain a better understanding, we conducted a large-
scale measurement study of the usage characteristics of on-
line social network based applications. In particular, we
developed and launched three Facebook applications, which
have achieved a combined subscription base of over 8 mil-
lion users. Using the rich dataset gathered through these
applications, we analyze the aggregate workload character-
istics (including temporal and geographical distributions) as
well as the structure of user interactions. We explore the
existence of ‘communities’, with high degree of interaction
within a community and limited interaction outside the com-
munity. We find that a small fraction of users account for
the majority of activity within the context of our Facebook
applications and a small number of applications account for
the majority of users on Facebook. Furthermore, user re-
sponse times for Facebook applications are independent of
source/destination user locality. We also investigate distin-
guishing characteristics of social gaming applications. To
the best of our knowledge, this is the first study analyzing
user activities on online social applications.
Categories and Subject Descriptors
C.2.0 [Computer - Communication Networks]: Gen-
eral; H.4.3 [Information Systems Applications]: Com-
munications Applications
General Terms
Measurement
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro t or commercial advantage an d that copies
bear this notice and the full citation on the rst page. To cop y otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci c
permission and/or a fee.
IMC’08, October 20 22, 2008, Vouliagmeni, Greece.
Copyright 2008 ACM 978-1-60558-334-1/08/10 ...$5.00.
Keywords
Online Social Networks, Social Games, Facebook, Applica-
tions, Characterization
1. INTRODUCTION
Over the past few years, online social networks (OSNs)
have attracted a massive following, with close to 90% of
undergraduate students in the United States using one or
the other social network on a regular basis [6]. As a result,
two OSNs (Facebook [21] and MySpace [28]) are now among
the top ten visited websites on the Internet [14].
OSNs have an inherent viral property in that applications’
user base can undergo exponential growth given the quick
spread of information much like real-world social networks.
Furthermore, through open developer platforms, large OSNs
such as Facebook and MySpace have recently opened their
doors to developers across the world, enabling even amateur
developers to create applications by leveraging the under-
lying social graphs. The introduction of these third-party
applications has led to even higher traffic on the correspond-
ing social networks. For example, there was 30% increase in
Facebook’s site traffic in the week following the launch of
its developer platform. Given the increasing popularity of
these applications, we believe it is important to characterize
such social network-based applications as a representative
modern class of workload.
This paper presents a detailed study of the usage char-
acteristics and nature of user interactions for three home-
grown applications launched using Facebook’s pioneering
Developer Platform1 [22]. We believe this is the first anal-
ysis of its kind. The key contributions of this paper are
summarized as follows:
• We developed and launched three applications using
the Facebook Developer Platform. Our applications
have been able to realize a combined user base of more
than 8 million users, placing them amidst the top 1%
of Facebook applications at the time of writing this pa-
per. We used these applications to procure a rich data
set on the usage of social network applications, which
has been made available to the Internet measurement
community2.
1We chose Facebook since it was the pioneer in launching its
Developer Platform (in May 2007). Moreover, multi-million
dollar investment and Facebook’s active development have
made Facebook Developer Platform the most evolved third-
party application base to date.
2Data available at http://www.ece.ucdavis.edu/rubinet/data.html

• We analyze various usage characteristics of our appli-
cations, such as geographical distribution of users, user
interactions and response times, and how they vary
with respect to the application type.
• We use our data set to infer the nature of user in-
teraction through Facebook applications. We model
this interaction through interaction graphs and show
that it exhibits small-world properties. One of our key
findings is that application dynamics can significantly
affect the structure of interaction graphs, hence weak-
ening the association between them and the underly-
ing real-world (friendship) relationships between users.
For example, user interaction graphs for non-gaming
applications are shown to contain stronger community
structures as compared to gaming applications.
• We also analyze global usage data for a broader set
of Facebook applications and show that application
popularity is characterized by a power-law distribu-
tion with exponential decay, and use our finding to
give insights into the underlying mechanism behind
application subscription and usage.
The paper is structured as follows. We begin with a brief
overview of the related work in Section 2. Section 3 describes
our data collection methodology and the design of our appli-
cations in detail. We then present high-level characteristics
of Facebook applications in Section 4, our findings regarding
community structures for our applications in Section 5, and
our findings related to user-level behavior for those appli-
cations in Section 6. We conclude with a discussion of our
results and future work in Section 7.
2. RELATED WORK
Over the past few years there has been a flurry of ac-
tivities on social network analysis. While some researchers
have focused on graph theoretic properties of social net-
works [7, 9, 10], others have analyzed individual networks’
usage patterns [2, 6]. However, there has not been a detailed
study of third-party applications developed and launched on
OSNs with a massive user base such as Facebook. We be-
lieve this paper is the first to measure and characterize this
new workload, the user interaction, and its relationship to
the underlying social networks.
Facebook has been the focus of a few studies recently. A
newly published study on characterization of Facebook ap-
plications [5] uses profile crawling to explore the high-level
characteristics of application users on Facebook, as well as
growth patterns of applications using publicly available us-
age statistics from Adonomics [16]. We confirm some of the
findings of this paper, and go beyond the scope of this study
by analyzing activity data from our home-grown applica-
tions.
Another important study by Golder et al. [6] on messaging
activity inside Facebook highlights Facebook-specific char-
acteristics such as regularities in daily and weekly traffic and
its relation to the use of Facebook by a select demographic
(college students). The same study found that activity on
Facebook seems to be focused on individual ‘networks’ and is
related to temporal usage patterns of those networks. Here,
‘networks’ refers to Facebook’s classification of users into
different networks of school, college, work and regional cat-
egories. We were able to confirm the findings of [6] with
regards to periodicity of traffic on Facebook, as well as ex-
tend our understanding of traffic patterns and user behavior
to third-party Facebook applications.
Other relevant studies include Newman’s work on com-
munity extraction algorithms [13] and Liben-Nowell’s work
on the relationship between geography and online friend-
ships [8]. We utilize results of the former and attempt to
extend Liben-Nowell’s findings by looking at user interaction
on social applications and its relation to users’ geographical
placement.
Furthermore, a recent study by Mislove et al. [10] focused
on the graph theoretic properties of large OSNs such as
YouTube [31], Flickr [25], and Orkut [29]. It discussed the
existence of small-world and scale-free properties. While we
do touch upon similar aspects in this study, note that we
focus on a new workload, namely third-party applications on
OSNs. In our study, we analyzed the actual user interactions
through our home-grown applications, rather than focusing
on the social networks determined through user friendship
profiles.
3. BACKGROUND AND METHODOLOGY
Facebook is a social networking website that has recently
gained immense popularity. Part of the reason for Face-
book’s success is its developer platform, which we shall dis-
cuss shortly. A friendship is formed on Facebook when one
Facebook user extends a (friendship) invitation to another
user. Upon confirmation by the latter, the friendship rela-
tionship is formed. Much of the activity on Facebook occurs
due to these friendship relationships. However, due to the
introduction of the Developer Platform, non-friend interac-
tions are now rising through interaction on social applica-
tions. Therefore, it is important to analyze users’ interac-
tions through these social applications, beyond the definition
of ‘friends’ through Facebook profiles.
In this section, we provide a brief overview of the Facebook
Developer Platform, followed by details of the applications
we implemented and a description of the data set used for
our study.
3.1 Facebook Developer Platform
The Facebook Developer Platform was launched in May
2007 [22] with little fanfare and only about eight applications
in its roster. Over the subsequent months, the Platform ex-
perienced phenomenal growth, showcasing more than 35,000
applications by July 2008 [16]. The launch of the Platform
also increased Facebook’s traffic by about 30% in the open-
ing week, and it has seen overall growth since [30].
Fig. 1 shows aspects of the Facebook Developer Plat-
form’s architecture that are relevant to our applications. In
this architecture, a user interacts indirectly with the appli-
cation servers through Facebook’s API servers. This enables
Facebook to protect users from malicious content that may
be embedded in the response data by the application servers,
since Facebook can process and strip undesirable content
from the server responses before forwarding them to users.
We must note, however, that Facebook has an alternate
method for deploying applications on its Platform that en-
ables users to interact directly with the application servers.