We provide a number of pioneering studies that explore various properties of popular applications on Facebook, which is the most populous Online Social Network (OSN) today. We begin with an analysis of the aggregate workload characteristics of social applications, as well as the structure of user interactions on social applications. We demonstrate the existence of communities in the applications' interaction graphs, and show that the amount of user activity on applications, and the number of users on applications, is power-law distributed. We also find that user response times for Facebook applications are independent of source/destination user locality.
We continue this dissertation with a study of the workload characteristics and performance of Facebook-based social applications from the network-level perspective. We present our measurement methodology that enabled us to gather, analyze, correlate, and report the workload characteristics of our popular social applications, and to measure performance from the perspective of the application servers. Coupled with PlanetLab experiments, where active probes are sent through Facebook to access a set of diverse applications, we study how Facebook forwarding/processing of requests/responses impacts the overall delay performance perceived by end-users.
We also employ data from our gifting applications to study the evolution of user activity through the most used growth mechanism on Facebook. We find user activity graphs differ from friendship graphs in terms of directionality of edges, transience of nodes, and asymmetry in in- and out-degrees distributions. We show that user activity graphs cannot be simulated through existing intent- and feature-driven algorithms that can model friendship graphs, and present a novel probabilistic growth model for user activity on gifting applications as a first step towards modeling activity from all genres of social applications.
Lastly, through a measurement-based study of a popular social game, we provide proof of the presence of phantom profiles (created solely to gain a strategic advantage) in social games. We show statistical differences among a subset of features associated with known genuine and phantom profiles. As a first step towards solving the more general problem of detecting phantom identities on OSNs, we then use supervised learning to classify phantom profiles.
Through this in-depth look at various aspects of social applications, we provide valuble insights to OSNs, application developers, and ISPs.
|Commitee:||Chuah, Chen-Nee, D'Souza, Raissa M., Wu, Felix|
|School:||University of California, Davis|
|School Location:||United States -- California|
|Source:||DAI-B 74/07(E), Dissertation Abstracts International|
|Keywords:||Algorithm, Gifting, Measurement, Modeling, Social applications, Social networks, User activities|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be