Abstract:
Facebook is the most popular Online Social Network (OSN) worldwide. It has grown from
one billion users in the third quarter of 2012 to two billions in the second quarter of 2017. The
usage of Facebook has developed from contacting friends to share personal information, and
it also became away for people to share, recommend, and link together all kinds of
information.
Facebook provides a huge set of behavioral data that if analyzed will produce different
kinds of useful information that could be benefited from in understanding human behavior in
depth, and impact design choices of social networks and applications. In addition to that,
analyzing these information help users to discover what they use Facebook for, and how much
time they spend using it. However, user behavior is not understood enough yet, because
accessing this type of data is limited.
By behavioral data we mean how the user interacts with Facebook functionalities. For
example, how many likes, comments, shares and other similar functions, in addition to the
content these functions applied to. In other words, we need to collect user interactions with
which content together with respective timing information. In particular, we focus on four
types of behavioral data: user’s actions, user’s friend list, activity log, and basic
demographical information. In this project we aim to update and redesign an outdated software
tools that were used in some previous similar studies [1]and [2].
Towards that end, we develop two software tools: First, Facebook Privacy Watcher and
Analyzer (FPWA) which is a Browser extension (plugin) -specifically for chrome-. Second,
Facebook Activity Log Collector (FALC) which is Android application for smartphones’
users. The plugin collects the data of participating users and stores it temporarily in the
Chrome browser storage. This enables us to make these data viewable by the user who can
then decide whether to send it or not to our server. In the same manner, the Android application
also collects user actions, but only the actions which stored in the activity log of Facebook
account.
We have a small number of users who participate in our project, according to that, the
analysis results won’t be comprehensive and representative. In spite of that, we develop a
Python program that analyzes any size of samples regardless of the amount of our collected
samples
